WO2012102015A1 - 画像データ処理装置、方法、プログラム及び集積回路 - Google Patents
画像データ処理装置、方法、プログラム及び集積回路 Download PDFInfo
- Publication number
- WO2012102015A1 WO2012102015A1 PCT/JP2012/000410 JP2012000410W WO2012102015A1 WO 2012102015 A1 WO2012102015 A1 WO 2012102015A1 JP 2012000410 W JP2012000410 W JP 2012000410W WO 2012102015 A1 WO2012102015 A1 WO 2012102015A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- face
- area
- feature
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
Definitions
- the present invention relates to an image data processing apparatus that calculates image feature information for classifying an image.
- Digital image capturing devices such as digital still cameras and mobile phones with camera functions have become widespread, and recording media such as hard disks for recording captured images have been provided at low cost.
- a user such as a digital image photographing apparatus (hereinafter simply referred to as a user) stores each photographed image in a recording medium such as a large capacity hard disk.
- a recording medium such as a large capacity hard disk.
- each image is classified into several classification categories It may be classified into For example, a case may be considered where an image taken at an athletic meet is classified into a classification destination category called an athletic meet.
- image feature information which is a feature of the image is calculated from each of the stored images, and the calculated image feature information is used.
- a technique for classifying each image into each classification destination category is known, and as described in Patent Document 2, a technique for classifying the image using the number and size of faces appearing in the image It has been known.
- each of the images is classified into a classification destination category to be originally classified. Can be difficult.
- the conventional image data processing apparatus calculates image feature information based on main colors (for example, black, blue, green, white, etc.) included in many images
- main colors for example, black, blue, green, white, etc.
- Images taken in the sea bathing often reflect the blue of the sea and white on the sand beach, and images taken in the ski trip often show sky blue and snow white in many cases. Therefore, the image feature information of the image captured in the swimming bath and the image feature information of the image captured in the ski trip, which are calculated by the image data processing apparatus, may be similar to each other.
- the present invention has been made in view of such problems, and is an image data processing apparatus for calculating image feature information for classifying an image, which uses the image feature information calculated from the image data processing apparatus.
- An image data processing apparatus for solving the above problems is an image data processing apparatus that calculates image feature information for classifying an image, and is a face specification that specifies a face area included in one image.
- an image feature calculation unit for calculating image feature information in the image from an image feature amount calculated based on at least a part of pixels of one image, the image feature calculation unit further comprising: The image feature quantity calculated based on the pixels included in the fixed area around the face area identified by the above is more than the image feature quantity calculated based on the pixels not included in the fixed area.
- the image feature information is calculated so as to be greatly reflected in the image feature information.
- a photographer of an image shoots an image including a human face at an event, it tends to shoot so that the feature of the event appears in an area around the face of the person. For example, in the sea bathing, the photographer tends to shoot an image so that the area around the person's face is blue-blue in the area around the person's face. In ski travel, the area around the person's face has snow white. There is a tendency for the image to be taken to be more.
- the image data processing apparatus having the above-described configuration is less likely to exhibit the feature of the event, the image feature value calculated based on the pixels of the area around the face of the person who tends to have the feature of the event.
- the image feature information is extracted by reflecting the image feature amount larger than the image feature amount calculated based on the pixels of the area away from the face of the person having a tendency.
- the image data processing apparatus can calculate image feature information in which the feature of the event is more largely reflected than the conventional image data processing apparatus.
- this image data processing apparatus can improve the classification accuracy in the case of classifying an image using the image feature information calculated from the image data processing apparatus, as compared with the conventional case.
- Functional block diagram showing the functional configuration of the image data processing apparatus 100 Directory structure diagram showing the directory structure of the image storage unit 231 A schematic diagram showing various areas visually A schematic diagram visually showing the area around people in special situations
- Data structure diagram of image feature information Data structure diagram of image family scene information
- Data structure diagram of image group family scene information Data structure of event feature information
- Flow chart of image feature information generation process Flow chart of image family scene information generation process
- Flow chart of image group family scene information generation process Flow chart of image group family scene information generation process Flow chart of image group classification process Diagram showing images taken at an event in which family members participate
- Functional block diagram showing the functional configuration of the image data processing apparatus 1700 Directory structure diagram showing the directory structure of the image storage unit 1731
- Data structure of event feature information Flow chart of deformed image feature information generation process
- Flow chart of image classification process Functional block diagram showing the functional configuration of the image data processing
- Embodiment 1 As one embodiment of the image data processing device according to the present invention, the human surrounding feature indicating the feature of the pixel around the person appearing in the image is calculated, and based on the calculated human surrounding feature, it is one event.
- An image data processing apparatus 100 that classifies an image into any one of a plurality of predetermined classification destination events which are different from each other in an image group unit consisting of a plurality of images photographed in association with a certain event. Will be explained.
- the image group is a set of images composed of a plurality of images designated by the user, and, for example, a set of images taken in an event of traveling to Hokkaido in winter 2009, for example, It is a set of images taken in the event of a trip to Okinawa in summer 2010.
- a classification destination event is a classification destination category that is a classification destination of an image group, and for example, a classification destination category that is a classification destination of an image group related to skis, for example, an image group related to sea bathing
- FIG. 1 is a hardware block diagram showing the main hardware configuration of the image data processing apparatus 100. As shown in FIG.
- the image data processing apparatus 100 communicates with a system LSI (Large Scale Integrated circuit) 110, a hard disk drive 130, an external recording medium read / write device 140, a USB control device 150, an output device 160, an input device 170, and
- the image processing apparatus 180 comprises an apparatus 180, and has a function of storing an image, which is a digital photograph, as data encoded by the JPEG (Joint Photographic Experts Group) method, and classifying the stored image.
- JPEG Joint Photographic Experts Group
- the image data processing apparatus 100 is connected to a device recording an image, represented by a digital still camera 192, via a detachable USB cable 195, and a display 193 and a monitor cable 196 for displaying an image.
- the system LSI 110 includes a CPU 101, a ROM 102, a RAM 103, a hard disk drive interface 104, an external recording medium read / write device interface 105, a USB (Universal Serial Bus) controller interface 106, an output device interface 107, and an input device. It is an LSI in which the interface 108, the communication device interface 109, the decoder 111, and the bus line 120 are integrated into one integrated circuit.
- the system LSI 110 is connected to the hard disk drive 130, the external recording medium read / write device 140, the USB control device 150, the output device 160, the input device 170, and the communication device 180.
- the CPU 101 is connected to the bus line 120, and executes a program stored in the ROM 102 or the RAM 103, whereby the ROM 102, the RAM 103, the hard disk drive 130, the external recording medium read / write device 140, the USB control device 150, the output device 160.
- the input device 170, the communication device 180, and the decoder 111 are controlled to realize various functions, for example, a function of reading image data stored in the hard disk device 130 into the memory area of the RAM 103, and the like.
- a function of reading out and decoding encoded image data from the image data and outputting the decoded image data to the display 193 is realized.
- the ROM 102 is connected to the bus line 120 and stores a program that defines the operation of the CPU 101 and data used by the CPU.
- the RAM 103 is connected to the bus line 120, temporarily stores data generated as the CPU 101 executes a program, and reads or writes data read from the hard disk drive 130 or the external recording medium reading / writing device 140. Data, data received by the communication device 180, data to be transmitted, etc. are temporarily stored.
- the decoder 111 is a DSP (Digital Signal Processor) having a function of decoding encoded image data, is connected to the bus line 120, is controlled by the CPU 101, and has a JPEG decoding function.
- DSP Digital Signal Processor
- the hard disk drive interface 104, the external recording medium reading and writing device interface 105, the USB control device interface 106, the output device interface 107, the input device interface 108, and the communication device interface 109 are respectively the hard disk drive 130 and the external recording medium reading and writing device
- An interface 140 mediates the exchange of signals between the USB control device 150, the output device 160, the input device 170, the communication device 180, and the bus line 120.
- the hard disk drive 130 is connected to the hard disk drive interface 104, is controlled by the CPU 101, and has a function of writing data in the built-in hard disk and a function of reading data written in the built-in hard disk.
- the image data is stored in a hard disk built in the hard disk device 130.
- the external recording medium reading / writing device 140 is connected to the external recording medium reading / writing device interface 105, is controlled by the CPU 101, has a function of writing data in the external recording medium, and a function of reading data written in the external recording medium And.
- the external recording medium is a DVD (Digital Versatile Disc), a DVD-R, a DVD-RAM, a BD (Blu-ray Disc), a BD-R, a BD-RE, an SD memory card 191, etc.
- the recording medium reading and writing apparatus 140 can read data from the DVD, BD, etc., and write and read data from the DVD-R, BD-R, BD-RE, SD memory card, etc. it can.
- the USB control device 150 is connected to the USB control device interface 106, is controlled by the CPU 101, has a function of writing data in an external device via a detachable USB cable 195, and a function of reading data written in the external device. Have.
- the external device is a device that stores an image, such as a digital still camera 192, a personal computer, a mobile phone with a camera function, and the USB control device 150 transmits the image to these external devices via the USB cable 195. It can write and read data.
- the output device 160 is connected to the output device interface 107 and the monitor cable 196, is controlled by the CPU 101, and has a function of outputting data to be displayed on the display 193 via the monitor cable 196.
- the input device 170 is connected to the input device interface 108, is controlled by the CPU 101, has a function of receiving an operation command from a user wirelessly transmitted from the remote control 197 and transmitting the received operation command to the CPU 101.
- the communication device 180 is connected to the communication device interface 109 and the network 194, is controlled by the CPU 101, and has a function of transmitting / receiving data to / from an external communication device via the network 194.
- the network 194 is realized by an optical communication line, a telephone line, a wireless line, and the like, and is connected to an external communication device, the Internet, and the like.
- the external communication device is a device such as an external hard disk drive or the like that stores an image, a program that defines the operation of the CPU 101, etc., and the communication device 180 transmits data from these external communication devices via the network 194. I can read it.
- the CPU 101 executes a program stored in the ROM 102 or the RAM 103, and the ROM 102, the RAM 103, the hard disk drive 130, the external recording medium reading / writing device 140, Various functions are realized by controlling the USB control device 150, the output device 160, the input device 170, the communication device 180, and the decoder 111.
- FIG. 2 is a functional block diagram showing the configuration of the main functional blocks of the image data processing apparatus 100. As shown in FIG.
- the image data processing apparatus 100 includes an image group data receiving unit 201, an image writing / reading unit 202, an image feature information writing / reading unit 203, a family scene information writing / reading unit 204, a face extraction unit 205, and a family scene information calculation unit 206.
- the image group data reception unit 201 is a block realized by the CPU 101 executing a program, and is connected to the image write / read unit 202, and an image of the image group 241 consisting of two or more images from the user. And the function to load the accepted image group into the memory area of the RAM 103 as an image group included in one image group, and the function to assign an image ID for specifying the image when the image is read And.
- an image may be read from an external communication device via the device 180.
- the image storage unit 231 is a storage area for storing a digital photograph as an image as image data encoded by the JPEG method, and is connected to the image writing / reading unit 202 and incorporated in the hard disk drive 130. Is implemented as part of the hard disk area.
- Each image data stored in the image storage unit 231 is logically managed by the directory structure under the file system as an image file.
- FIG. 3 is a directory structure diagram showing the directory structure of the image storage unit 231. As shown in FIG. 3
- the directory structure of the image storage unit 231 is composed of a total of three layers of the highest hierarchy 310, the first directory hierarchy 320, and the second directory hierarchy 330.
- the first directory hierarchy 320 there are a plurality of classified destination event directories such as a ski directory 321, a swimming directory 322, a picnic directory 323, and an actual data storage directory 324.
- the classification destination event directory is a directory having the same name as the classification destination event which is the classification destination of the image group, and there is only one directory of the same name.
- the actual data storage directory 324 is a directory for storing image data, and the data of the image is stored only in the actual data storage directory 324.
- the second directory hierarchy 330 there are a plurality of event directories such as the 2010 winter Shinshu travel directory 331, the 2009 winter Hokkaido travel directory 332, and the summer 2010 Okinawa travel directory.
- the event directory is a directory corresponding to an image group consisting of an image group accepted by the image group data accepting unit 201, and among the data held in the actual data storage directory 324, all the images belonging to the image group By holding information indicating the address of data, it is a directory in which data of the image is linked.
- Each event directory exists under the classified event directory corresponding to the classified event into which the corresponding image group is classified.
- the name of each event directory is an event name designated by the user using the image data processing apparatus 100 for the image group corresponding to the event directory.
- the method of generating each event directory will be described later in ⁇ Image Group Classification Processing>.
- the image writing / reading unit 202 is a block realized by the CPU 101 executing a program, and the image storage unit 231, the image group data receiving unit 201, the face extracting unit 205, and the family scene information calculating unit 206. And a function of reading out the image stored in the image storage unit 231, connected to the image group classification unit 208, a function of writing the image in the image storage unit 231, and a function of changing the directory structure of the image storage unit 231 , And a function of changing the link of the image data of the image storage unit 231.
- the sample image storage unit 236 is a storage area for storing a sample image, which is a digital photograph in which the face of a specific person (for example, a family) is shown, as image data encoded by the JPEG method. It is connected to the writing unit 214 and the face extracting unit 205, and is implemented as a partial area of a hard disk built in the hard disk device 130.
- the sample image writing unit 214 is a block realized by the CPU 101 executing a program, and is connected to the sample image receiving unit 213 to specify a sample image and a person received by the sample image receiving unit 213. It has a function of writing the face ID into the sample image storage unit 236.
- the sample image receiving unit 213 is a block realized by the CPU 101 executing a program, and is connected to the sample image writing unit 214, and a sample image in which a user's face of a specific person is shown and the person And the function of reading the received sample image and the corresponding face in correspondence with each other and reading the received sample image into the memory area of the RAM 103 and the read sample image using the sample image writing unit 214 It has a function to be stored in the storage unit 236.
- an image may be read from an external communication device via 180.
- the face extraction unit 205 is a block realized by the CPU 101 executing a program, and is connected to the image writing / reading unit 202, the human surrounding feature extraction unit 207, and the sample image storage unit 236, and Have three functions.
- Function 1 When a predetermined face model indicating human face features is held and face recognition is attempted by referring to the held face model and face recognition is performed. The function of calculating the area of the recognized face area and the position of the recognized face, and sequentially assigning face IDs for identifying the recognized face to each of the recognized faces.
- the face model is, for example, the brightness of parts forming the face, such as eyes, nose, and mouth, information on relative positional relationship, etc.
- the recognized face area is, for example, recognized Among the rectangles including the sides in the horizontal direction in the image and the sides in the vertical direction in the image including the face, it is a rectangular area with the smallest area.
- the face extraction unit 205 uses, for example, the one held in the hard disk drive 130 as a predetermined face model, it is conceivable that the face extraction unit 205 refers to one stored outside.
- Function 2 When the face is recognized, the feature of the recognized face and the feature of the face included in the sample image stored in the sample image storage unit 236 are extracted from the image, and the sample image storage unit 236 is extracted. When there is an image having the same face feature as the recognized face feature among the stored sample images, it is assumed that the person of the recognized face is the same person as the person shown in the sample image. Function to judge.
- the feature of the face means, for example, a relative positional relationship of parts forming the face such as eyes, a nose, and a mouth, an area ratio of these parts, and the like.
- Function 3 When a face is recognized, the body area is calculated below the recognized face area as a rectangular area defined by a predetermined algorithm with respect to the recognized face area. The function of calculating the area of the area of and the position of the calculated area of the body, and sequentially assigning the calculated body ID to each of the calculated areas of the body to specify the calculated area of the body.
- the predetermined algorithm for determining the area of the body is, for example, 1.5 times the width in the horizontal direction in the image of the face area below the recognized face area and the vertical direction in the image of the face area
- a rectangular area whose width is doubled and whose horizontal coordinate in the image of the center point matches the horizontal coordinate in the image of the center of the face area is the body area. It is an algorithm.
- the human feature extraction unit 207 is a block realized by the CPU 101 executing a program, and is connected to the face extraction unit 205 and the image feature information write / read unit 203, and has the following five functions: Have.
- Function 1 A function of calculating the position of a face surrounding area calculated as a rectangular area determined by a predetermined algorithm with respect to the position of the face area calculated by the face extracting unit 205.
- the predetermined algorithm for determining the face surrounding area is, for example, adding the horizontal width in the image of the face area to the left and right in the horizontal direction in the image with respect to the face area, and Above the direction, a rectangular area obtained by adding the width in the vertical direction of the image to the area of the face is used as the area around the face.
- Function 2 A function of calculating the position of the body surrounding area by calculating the body surrounding area as a rectangular area defined by a predetermined algorithm with respect to the position of the face area calculated by the face extracting unit 205.
- the predetermined algorithm for determining the area around the body is, for example, adding the horizontal width in the image of the face area to the left and right in the horizontal direction in the image with respect to the body area, and This is an algorithm that sets a rectangular area obtained by adding a half width of the vertical width in the image to the upper and lower sides of the direction as the body peripheral area.
- Function 3 A function of calculating a human surrounding area as an area determined by a predetermined algorithm for the calculated face peripheral area and the body peripheral area.
- the predetermined algorithm for determining the human surrounding area is, for example, a human peripheral area excluding an area in which the face area and the body area are excluded from the area included in at least one of the face peripheral area and the body peripheral area. It is an algorithm that
- Function 4 For each pixel included in the human-peripheral area, the color of the pixel is determined in advance from the color components constituting the pixel, for example, the luminance values of R (Red), G (Green), and B (Blue). Identify which one of the N major colors (eg, black, blue, green, white, etc.) that are defined is similar, and for each of the identified colors, the total area included in the human-peripheral area A function of calculating the ratio of the number of pixels specified for the color to the number of pixels as a human feature amount.
- a method of identifying to which main color the color of a certain pixel is similar for example, the range of the luminance value of R, the range of the luminance value of G, and the range of B corresponding to each of the main colors in advance
- a method of specifying which main color the pixel is similar to by specifying the range of the luminance value and comparing with the luminance value of R, G, B of the pixel to be specified.
- Function 5 Function to generate image feature information (described later).
- FIG. 4 is a schematic view visually showing the various regions described above.
- an image 401 is an image in which a person 411 including a face 412 and a body 413 is photographed.
- the first deformed image 402 a region 422 of the face calculated by the face extraction unit 205 in the image 401 is shown.
- the region 423 of the body calculated by the face extraction unit 205 in the image 401 is shown.
- the third deformed image 404 shows a face surrounding area 424 calculated by the human surrounding feature quantity extraction unit 207 in the image 401.
- a body peripheral region 425 calculated by the human peripheral feature extraction unit 207 in the image 401 is shown.
- a human-surrounding area 426 calculated by the human-surrounding feature extraction unit 207 in the image 401 is shown.
- the area 422 of the face, the area 423 of the body, the area around the face 424, the area around the body 425, and the area around the person 426 are calculated.
- FIG. 5 is a schematic view visually showing the area around a person in a special situation.
- an image 501 is an example of an image including a plurality of recognized faces, and here, 2 of a person 511 consisting of a face A 512 and a body A 513 and a person 561 consisting of a face B 562 and a body B 563 It is an image that includes the recognized face of a person.
- a human-surrounding area 526 calculated by the human-surrounding feature extraction unit 207 in the image 501 is shown.
- the human surrounding feature quantity extraction unit 207 corresponds to the face surrounding area corresponding to the face A512, the face surrounding area corresponding to the face B562, the body surrounding area corresponding to the body A513, and the body B563. From the region included in at least one of the peripheral region, the region of the face corresponding to face A512, the region of the face corresponding to face B562, the region of the body corresponding to body A513, and the region of the body corresponding to body B563 The area excluding [1] is calculated as the area around people.
- the human surrounding feature amount extraction unit 207 performs at least one of any one of the face surrounding area and any of the body surrounding areas. An area in which all face areas and all body areas are excluded from the areas included in is taken as the area around people.
- the image 503 is an example of an image including a recognized face in which a part of the face surrounding area or a part of the body surrounding area protrudes from the image, and here, a person 591 including a face 592 and a body 593 It is an included image.
- a human-surrounding area 596 calculated by the human-surrounding feature extraction unit 207 in the image 503 is shown.
- the human surrounding feature quantity extraction unit 207 extracts the face area and the body area from the area included in at least one of the face surrounding area and the body surrounding area among the areas in the image.
- An area excluding the area not included in the image 503 is calculated as human surrounding information.
- the image feature information storage unit 232 is a storage area for storing image feature information, is connected to the image feature information write / read unit 203, and is implemented as a partial area of a hard disk built in the hard disk drive 130. It is done.
- FIG. 6 is a view showing an example of the data structure of the image feature information stored in the image feature information storage unit 232. As shown in FIG.
- an image ID 600 As shown in the figure, in the image feature information, an image ID 600, a face feature amount 610, a body feature amount 620, a face surrounding area 630, a body around area 640, and a person surrounding feature amount 650 are associated. Is configured.
- the face feature amount 610 further includes a face ID 611, a face area 612, a face position 613, and a corresponding face ID 614
- the body feature amount 620 further includes a body ID 621 and a body area. 622 and the body position 623 are associated with each other, and the human surrounding feature 650 is further associated with the black ratio 651, the blue ratio 652, the green ratio 653, and the white ratio 654. ing.
- the image ID 600 is an ID for identifying an image, which is given to each image by the image group data receiving unit 201.
- the face ID 611 is an ID for identifying the recognized face given to each face recognized by the face extraction unit 205.
- the face area 612 is a ratio of the area of the face area to the area of the image calculated by the face extraction unit 205, and is normalized to 1 when the face area is the entire image.
- the face position 613 is coordinates indicating the position of the face area calculated by the face extraction unit 205, and is composed of the upper left coordinates and the lower right coordinates of the rectangular face area in the image.
- the corresponding face ID 614 is an ID for specifying a person included in the image stored in the sample image storage unit 236.
- the corresponding face ID 614 is “01”, the person is a son, “02” is a father, and “03” is a mother.
- the corresponding face ID 614 is "99".
- the corresponding face ID 614 is “01” or “02” or “03”, it is assumed that the person is a family.
- the body ID 621 is an ID for identifying the body area, which is given to the body area calculated by the face extraction unit 205.
- the body area 622 is a ratio of the area of the body area to the area of the image calculated by the face extraction unit 205, and is normalized to 1 when the body area is the entire image.
- the body position 623 is coordinates indicating the position of the body area calculated by the face extraction unit 205, and is composed of the upper left coordinates and the lower right coordinates of the rectangular body area in the image.
- the face surrounding area 630 is coordinates indicating the position of the face surrounding area calculated by the human surrounding feature extraction unit 207, and is configured by the upper left coordinates and the lower right coordinates of the face surrounding area in the image. .
- the coordinates are X, Y coordinates when the upper left coordinate of the image is (0, 0).
- the body peripheral area 640 is coordinates indicating the position of the body peripheral area calculated by the human body feature extraction unit 207, and is the coordinates of the upper left and lower right coordinates of the rectangular body peripheral area in the image. Configured
- the black ratio 651 is a ratio of the number of pixels specified as black contained in the human peripheral area to the total number of pixels contained in the human peripheral area, calculated by the human peripheral feature quantity extraction unit 207.
- the blue ratio 652 is a ratio of the number of pixels specified as blue contained in the human surrounding area to the total number of pixels contained in the human surrounding area, calculated by the human surrounding feature extraction unit 207.
- the green ratio 653 is a ratio of the number of pixels specified as green included in the human surrounding area to the total number of pixels included in the human surrounding area, calculated by the human surrounding feature extraction unit 207.
- the white ratio 654 is a ratio of the number of pixels specified as white contained in the human peripheral area, calculated by the human peripheral feature quantity extraction unit 207, to the total number of pixels contained in the human peripheral area.
- the image feature information writing and reading unit 203 is a block realized by the CPU 101 executing a program, and includes a human surrounding feature amount extracting unit 207, a family scene information calculating unit 206, and an image feature information storage unit 232. , And has a function of reading and writing image feature information from and to the image feature information storage unit 232.
- the event name information reception unit 211 is a block realized by the CPU 101 executing a program, and is an image group connected to the family scene information calculation unit 206 and input by a user who uses the image data processing apparatus 100. It has a function to receive the event name which is the name of.
- Family scene information calculation unit 206 is a block realized by CPU 101 executing a program, and image write / read unit 202, image feature information write / read unit 203, and family scene information write / read unit 204. And the event name information receiving unit 211, and has the following two functions.
- Function 1 A function of calculating an image family scene feature (described later) for an image including a recognized face, as a value determined by a predetermined algorithm with respect to a person surrounding feature.
- each image family scene color ratio of the image family scene feature value is obtained by dividing the value of each color ratio of the human surrounding feature value by the value of the face area.
- the calculated image family scene feature quantity is weighted such that the value of the image family scene feature quantity of the image becomes larger in the image having a smaller area of the face area in the image. It will be done.
- Function 2 A function of calculating an image group family scene feature (described later) for an image group as a value determined by a predetermined algorithm with respect to an image family scene feature of an image included in the image group.
- the predetermined algorithm for determining the image group family scene feature amount is, for example, each image of the image family scene feature amount corresponding to an image including a face recognized as a family face among images included in the image group It is an algorithm which makes the average value about the value of a family scene color ratio the value of each image group family scene color ratio of an image group family scene feature-value.
- the face of a family is a face in which the corresponding face ID indicates a family.
- Function 3 A function to generate image family scene information (described later) and image group family scene information (described later).
- Family scene information storage unit 233 is a storage area for storing image family scene information and image group family scene information, and is connected to family scene information write / read unit 204 and is incorporated in hard disk drive 130. It is implemented as part of the hard disk area.
- FIG. 7 is a view showing an example of the data structure of image family scene information stored in the family scene information storage unit 233. As shown in FIG.
- an image ID 700, a face ID 710, a corresponding face ID 720, and an image family scene feature value 730 are associated with each other.
- the image family scene feature value 730 is further configured by correlating the image family scene black ratio 731, the image family scene blue ratio 732, the image family scene green ratio 733 and the image family scene white ratio 734. ing.
- the image ID 700, the face ID 710, and the corresponding face ID 720 are equivalent to the image ID 600, the face ID 611, and the corresponding face ID 614 in FIG. 6, respectively. Therefore, the description is omitted.
- the image family scene black ratio 731 is a value calculated by weighting the value of the black ratio 651 (see FIG. 6) of the corresponding image, and the family scene information calculation unit 206 calculates the value of the black ratio 651. Is calculated by dividing the value by the value of the corresponding face area 612.
- the image family scene blue ratio 732 is a value calculated by weighting the value of the blue ratio 652 of the corresponding image, and the family scene information calculation unit 206 sets the value of the blue ratio 652 to the corresponding face. It is calculated by dividing by the value of the area 612.
- the image family scene green ratio 733 is a value calculated by weighting the value of the green ratio 653 of the corresponding image, and the family scene information calculation unit 206 sets the value of the green ratio 653 to the corresponding face. It is calculated by dividing by the value of the area 612.
- the image family scene white ratio 734 is a value calculated by weighting the value of the white ratio 654 of the corresponding image, and the family scene information calculation unit 206 sets the value of the white ratio 654 to the corresponding face. It is calculated by dividing by the value of the area 612.
- FIG. 8 is a view showing an example of the data structure of the image group family scene information stored in the family scene information storage unit 233. As shown in FIG.
- the image group family scene information is configured by associating an image group ID 800, an event name 810, and an image group family scene feature value 820 with each other.
- the image group family scene feature value 820 further corresponds to an image group family scene black ratio 821, an image group family scene blue ratio 822, an image group family scene green ratio 823, and an image group family scene white ratio 824. It is attached and configured.
- the image group ID 800 is an ID for specifying an image group.
- the event name 810 is an event name, which is the name of an image group, which is input by the user using the image data processing apparatus 100 via the event name information receiving unit 211.
- the image group family scene black ratio 821 is an average value of the values of the image family scene black ratio 731 (see FIG. 7) in the image including the face recognized as the family face among the images constituting the corresponding image group. It is calculated by the family scene information calculation unit 206.
- the image group family scene blue ratio 822 is an average value of the values of the image family scene blue ratio 732 in the image including the face recognized as the face of the family among the images constituting the corresponding image group, and the family scene information It is calculated by the calculation unit 206.
- the image group family scene green ratio 823 is an average value of the values of the image family scene green ratio 733 in the image including the face recognized as the face of the family among the images constituting the corresponding image group, and the family scene information It is calculated by the calculation unit 206.
- the image group family scene white ratio 824 is an average value of the values of the image family scene white ratio 734 in the image including the recognized face of the family among the images constituting the corresponding image group, and the family scene information calculation unit It is calculated by 206.
- the family scene information write / read unit 204 is a block realized by the CPU 101 executing a program, and is connected to the family scene information calculation unit 206, the image group classification unit 208, and the family scene information storage unit 233. It has a function of reading and writing image family scene information and image group family scene information to the family scene information storage unit 233.
- the event feature information storage unit 234 is a storage area for storing event feature information, is connected to the event feature information write / read unit 209, and is implemented as a partial area of a hard disk built in the hard disk drive 130. It is done.
- FIG. 9 is a view showing an example of the data structure of event feature information stored in the event feature information storage unit 234. As shown in FIG.
- the event feature information includes an image group family scene black ratio 1.5 or more, an image group family scene blue ratio 1.5 or more 902, an image group family scene ratio green 1.5 or more 903, an image
- classification conditions 900 such as group family scene ratio white 1.5 or more 904 is associated with each of classification destination events 910 such as fireworks 911, sea bath 912, picnic 913, ski 914 and the like.
- the event feature information write / read unit 209 is a block realized by the CPU 101 executing a program, and is connected to the image group classification unit 208, the event feature information reception unit 212, and the event feature information storage unit 234.
- the event feature information storage unit 234 has a function of reading and writing event feature information.
- the event feature information reception unit 212 is a block realized by the CPU 101 executing a program, and is input by a user who connects the event feature information write / read unit 209 and uses the image data processing apparatus 100. It has a function of receiving event feature information and storing the received event feature information in the event feature information storage unit 234 using the event feature information write / read unit 209.
- the event control information reception unit 212 receives the event characteristic information from the external recording medium via the external recording medium reading / writing device 140, and the USB control device 150. There are a case of receiving from an external device via the connection and a case of receiving from an external communication device via the communication device 180.
- the image group classification unit 208 is a block realized by the CPU 101 executing a program, and includes an image write / read unit 202, a family scene information write / read unit 204, and an event feature information write / read unit 209. , Image group based on the image group family scene information stored in the family scene information storage unit 233 and connected to the classification result output unit 210 and the event feature information stored in the event feature information storage unit 234 Are classified into classified events.
- the classification result output unit 210 is a block realized by the CPU 101 executing a program, and is connected to the image group classification unit 208, and displays the classification result when the image group classification unit 208 classifies the image groups. It has a function to display at 193.
- the characteristic operations performed by the image data processing apparatus 100 include image feature information generation processing, image family scene information generation processing, image group family scene information generation processing, and image group classification processing.
- the image feature information generation process is a process in which the image data processing apparatus 100 reads an image in image group units and generates image feature information for each of the read images.
- FIG. 10 is a flowchart of image feature information generation processing performed by the image data processing apparatus 100.
- the image feature information generation process is started when the remote control 197 receives an operation from the user to start the image feature information generation process.
- the image group data receiving unit 201 starts reading an image of one image group, and the event name information receiving unit 211 generates an event in which an image belonging to the image group is captured.
- the reception of the event name is started (step S1000).
- the image group data reception unit 201 can use an external recording medium mounted on the external recording medium reading / writing device 140 or an external device via the USB cable 195 connected to the USB control device 150 or the network 194. Images can be read from the connected communication device 180.
- the image group data reception unit 201 reads an image recorded in the SD memory card 191 one by one, sequentially assigns an image ID to the read image, and associates the image data with the image ID.
- the image writing and reading unit 202 writes the actual data storage directory 324 of the image storage unit 231 using the image writing and reading unit 202.
- the event name information receiving unit 211 receives an event name of an event in which an image belonging to an image group is photographed by the operation of the remote control 197 from the user.
- face extraction unit 205 receives the image group accepted by image group data acceptance unit 201 from image storage unit 231 using image write / read unit 202. Is selected and read out one by one (step S1010), and the read out image is decoded by the JPEG method.
- the face extraction unit 205 tries to recognize the face included in one image by referring to the model of the face to be held for one read image (step S1020).
- the face extraction unit 205 calculates face feature amounts for each of the recognized faces (Step S1030). That is, the area of the recognized face area and the position of the recognized face are calculated, and the face ID for identifying the recognized face is sequentially assigned to each recognized face, and the recognized face is identified.
- the feature and the feature of the face included in the sample image stored in the sample image storage unit 236 are extracted, and the sample image stored in the sample image storage unit 236 is the same as the recognized feature of the face
- the corresponding face ID corresponding to the sample image is assigned to the corresponding face ID of the recognized face. If there is no image having the same facial feature as the recognized facial feature in the sample image stored in the sample image storage unit 236, it is indicated that the corresponding face ID of the recognized face is another person Grant "99" which means
- the face extraction unit 205 calculates a body region below the recognized face region for each of the recognized faces, and calculates a body feature value for each of the calculated body regions. (Step S1040). That is, the area of the calculated body area and the position of the calculated body area are calculated, and body IDs for specifying the calculated body area are sequentially assigned to each of the calculated body areas. Do.
- the human surrounding feature quantity extraction unit 207 calculates a face surrounding area based on the face area, calculates a body surrounding area based on the body area, and the calculated face surrounding area.
- the human-peripheral area is calculated from the calculated per-body area (step S1050).
- the human surrounding feature quantity extraction unit 207 calculates a human surrounding feature quantity based on the pixel values of the pixels included in the human surrounding area (step S1060). That is, for each pixel included in the human surrounding area, the color of the pixel is specified from each luminance of R, G, and B which are color components constituting the pixel, and for each specified color, the human surrounding area The ratio of the number of pixels specified for that color to the total number of pixels contained in is calculated as the color ratio of that color.
- step S1020 when the face extraction unit 205 does not recognize the face (step S1020: No), the face extraction unit 205 sets a null value as the value of each component of the face feature amount, and the body A null value is set as the value of each component of the feature quantity, the human surrounding feature quantity extraction unit 207 sets a null value as the value of the face surrounding area, and a null value as the value of the body surrounding area. A null value is set as the value of each component of the feature amount (step S1070).
- the human surrounding feature quantity extraction unit 207 When the process of step S1060 is completed, or when the process of step S1070 is completed, the human surrounding feature quantity extraction unit 207 generates image feature information for the target image, and generates the generated image feature information.
- the image feature information storage unit 232 stores the image feature information using the image feature information writing / reading unit 203 (step S1080).
- step S1080 the face extraction unit 205 checks whether there is an image not yet selected among the images belonging to the image group received by the image group data reception unit 201 (step S1090). .
- step S1090 when there is an unselected image (step S1090: No), the image data processing apparatus 100 returns to the process of step S1010 again and continues the process of step S1010 and subsequent steps.
- step S1090 when there is no unselected image (step S1090: YES), the image data processing apparatus 100 ends the image feature information generation process.
- the image family scene information generation process is a process in which the image data processing apparatus 100 generates image family scene information based on the image feature information.
- FIG. 11 is a flowchart of image family scene information generation processing performed by the image data processing apparatus 100.
- the image family scene information generation process is started when the image data processing device 100 ends the image feature information generation process.
- the family scene information calculation unit 206 uses the image feature information write / read unit 203 to be processed in the image feature information generation process from the image feature information storage unit 232. Image feature information of an image belonging to the selected image group is read out (step S1100).
- the family scene information calculation unit 206 selects one piece of image feature information from the read out image feature information (step S1110), and the recognized face is included in the image corresponding to the selected image feature information. It is checked (step S1120). Here, whether or not the recognized face is included is checked by checking whether each component constituting the face feature value is a null value or not.
- the family scene information calculation unit 206 calculates the image family scene feature amount from the face area 612 (see FIG. 6) and the human periphery feature amount 650. (Step S1130). That is, by dividing each of the values of the respective color ratios constituting the human surrounding feature amount 650 by the value of the face area 612, the values of the respective image family scene color ratios are calculated.
- the family scene information calculation unit 206 sets a null value as the value of the image family scene feature value (step S1140). . That is, the values of the respective image family scene color ratios constituting the image family scene feature value are set as null values.
- step S1130 When the process of step S1130 is completed, or when the process of step S1140 is completed, the family scene information calculation unit 206 generates image family scene information for the image feature information that is the object, and the generated image family The scene information is stored in the family scene information storage unit 233 using the family scene information write / read unit 204 (step S1150).
- the family scene information calculation unit 206 checks whether or not there is image feature information that has not been selected yet in the image feature information of the images belonging to the image group to be targeted (step S1160). ).
- step S1160 when there is unselected image feature information (step S1160: No), the image data processing apparatus 100 returns to the process of step S1110 again and continues the process of step S1110 and subsequent steps.
- step S1160 when there is no unselected image feature information (step S1160: Yes), the image data processing apparatus 100 ends the image family scene information generation process.
- the image group family scene information generation process is a process in which the image data processing apparatus 100 generates image group family scene information of the image group based on the image family scene information of each image belonging to the image group.
- FIG. 12 is a flowchart of an image group family scene information generation process performed by the image data processing apparatus 100.
- the image group family scene information generation process is started when the image data processing apparatus 100 ends the image family scene information generation process.
- the family scene information calculation unit 206 uses the family scene information write / read unit 204 to set the target in the image family scene information generation process from the family scene information storage unit 233.
- Image family scene information of an image belonging to the image group which has become absent is read out (step S1200).
- the family scene information calculation unit 206 calculates, as an image group family scene feature quantity, an average value of image family scene feature quantities for the corresponding face ID 720 (see FIG. 7) indicating a family among the read out image family scene information. (Step S1210). That is, for the image family scene information in which the corresponding face ID 720 indicates a family, the average value of each image family scene color ratio constituting the image family scene feature value 730 is calculated to obtain each image group family scene color ratio. Calculate the value.
- the family scene information calculation unit 206 calculates the value of each image group family scene color ratio constituting the image group family scene feature value, Each has a null value.
- the family scene information calculation unit 206 When the process of step S1210 is completed, the family scene information calculation unit 206 generates image group family scene information for the target image group, and the generated image group family scene information is stored in the family scene information write / read unit.
- the image data processing apparatus 100 stores the image data in the family scene information storage unit 233 (step S 1220) using the information 204, and ends the image group family scene information generation processing.
- the event name 810 (see FIG. 8) is the event name accepted from the user by the event name information accepting unit 211 in step S1000 of the image feature information creation process.
- the image group classification process is a process in which the image data processing apparatus 100 classifies an image group into any of classification destination events.
- FIG. 13 is a flowchart of image group classification processing performed by the image data processing apparatus 100.
- the image group classification process is started when the image data processing apparatus 100 ends the image group family scene information generation process.
- the image group classification unit 208 uses the family scene information writing and reading unit 204 to send the image group family scene information generation process from the family scene information storage unit 233.
- the image group family scene information belonging to the group is read out, and the event characteristic information is read out from the event characteristic information storage unit 234 using the event characteristic information write / read unit 209 (step S1300).
- the image group classification unit 208 compares the read image group family scene information with the read event feature information (step S1310), and the classification destination event of the image group Is calculated (step S1320). That is, it is checked whether there is any image group family scene color ratio (see FIG. 8) included in the image group family scene information that matches the classification condition 900 (see FIG. 9) included in the event feature information.
- the classification destination event 910 corresponding to the matching classification condition 900 is calculated as the classification destination event to which the image group should be classified.
- the classification destination event "other event" is calculated as the classification destination event to which the image group is to be classified.
- a classification destination event called another event is calculated as a classification destination event to which the image group is to be classified.
- all matching destination events 910 are calculated as the destination events to which the image group should be assigned.
- the image group classification unit 208 uses the image writing and reading unit 202 to display an image group under the classification destination event directory corresponding to the classification destination event to be classified in the image storage unit 231. Create an event directory with the same name as the event name associated with the event directory, hold information indicating the address of data of all the images belonging to the image group under the event directory, and belong to the image group The image groups are classified by setting the data of all the images to be linked (step S1330).
- the classification result output unit 210 causes the display 193 to display the classification destination event name of the classification destination event to be classified calculated by the image group classification unit 208 together with the event name associated with the image group.
- the data processing apparatus 100 ends the image group classification process.
- FIG. 14 is an example of an image group captured at an event in which family members participate.
- An image group 1400 is an image group consisting of images taken in a family ski trip, and is composed of an image 1401 and an image 1402, and the associated event name is “2010 Toshinshu Trip”. It has become.
- the image 1401 includes a son who enjoys skiing, and the image 1402 includes a father who enjoys skiing. These images are rich in sky blue and snow white, but they are characterized by a large amount of snow white symbolizing an event of skiing around the person.
- the image feature information of the image 1401 corresponds to, for example, “0001” in the image ID 600 in FIG. 6, and the image family scene information in the image 1401 corresponds to, for example, “0001” in the image ID 700 in FIG. It is assumed that
- the image feature information of the image 1402 corresponds to, for example, the image ID 600 in FIG. 6 corresponding to “0002”, and the image family scene information of the image 1402 corresponds to, for example, the image ID 700 in FIG. It is assumed that
- the image group ID 800 in FIG. 8 corresponds to "001".
- An image group 1410 is an image group consisting of an image group taken in a sea bathing trip conducted by a family, and is composed of an image 1411 and an image 1412, and the associated event name is "2010 Okinawa travel". It has become.
- the image 1411 includes a son enjoying sea bathing, and the image 1412 includes a father and a mother enjoying sea bathing. These images are rich in the blue color of the sea and the white color of the sandy beach, but they are characterized in that the blue color of the sea, which symbolizes the event of sea bathing, is often contained around the person.
- the image feature information of the image 1411 corresponds to, for example, the image ID 600 in FIG. 6 corresponding to “0003”, and the image family scene information of the image 1411 corresponds to, for example, the image ID 700 in FIG. It is assumed that
- the image feature information of the image 1412 corresponds to, for example, “0004” in the image ID 600 in FIG. 6, and the image family scene information in the image 1412 corresponds to, for example, “0004” in the image ID 700 in FIG. It is assumed that
- the image group ID 800 in FIG. 8 corresponds to “002”.
- the image data processing apparatus 100 can set the feature of the image to the area around the person even if the respective images belonging to different image groups are similar to each other in the features of the entire image. By extracting from the surrounding area, these image groups can be classified into different classified events.
- a conventional image data processing apparatus extracts feature of an image from the entire image, and assuming that a feature amount indicating a feature of an image belonging to an image group is called an image group scene feature amount, an image group scene feature amount of the image group 1400 And the image group scene feature quantities of the image group 1410 are similar to each other.
- FIG. 15 is a view showing an example of the data structure of image group scene information generated by the conventional image data processing apparatus.
- the image group scene information is configured by associating an image group ID 1500, an event name 1510, and an image group scene feature quantity 1520.
- the image group scene feature quantity 1520 further includes an image group scene black ratio 1521, an image group scene blue ratio 1522, an image group scene green ratio 1523, and an image group scene white ratio 1524 in association with each other. ing.
- An image group ID 1500 is an ID for specifying an image group.
- the image group whose image group ID 1500 is “001” is the image group 1400, and the image group whose image group ID 1500 is “002” is the image group 1410.
- the event name 1510 is an event name which is a name of an image group.
- the image group scene black ratio 1521 is an average value of the ratio of the number of pixels identified as black to the number of pixels of the entire image in each of the images constituting the corresponding image group.
- the image group scene blue ratio 1522 is an average value of the ratio of the number of pixels identified as blue to the number of pixels of the entire image in each of the images constituting the corresponding image group.
- the image group scene green ratio 1523 is an average value of the ratio of the number of pixels identified as green to the number of pixels of the entire image in each of the images constituting the corresponding image group.
- the image group scene white ratio 1524 is an average value of the ratio of the number of pixels identified as white to the number of pixels of the entire image in each of the images constituting the corresponding image group.
- FIG. 16 is a diagram showing two images.
- An image 1601 is an image taken in a family ski trip, and includes a son who enjoys skiing. This image is an image taken with awareness of the event of a ski trip conducted by a family, and taken so as to include the whole body of the son against the background of the snowy mountains.
- An image 1602 is an image taken in the city, and is an image of a close-up of a mother's face including a large mother's face. This image is an image photographed with awareness of the mother's face, and is photographed without particular attention to the background.
- each value of each image family scene color ratio of image family scene feature values is calculated by dividing each color ratio value of human surrounding feature values by the value of the corresponding face area. And thereby, the calculated image family scene feature quantity is weighted such that the value of the image family scene feature quantity of the image becomes larger in the image having a smaller area of the face area in the image. It has become.
- the image data processing apparatus 1700 has the same hardware configuration as the image data processing apparatus 100 according to the first embodiment, but a part of the program to be executed is the image data processing apparatus according to the first embodiment. It is different from 100.
- the image data processing apparatus 100 according to the first embodiment is an example of classifying images in image group units, but the image data processing apparatus 1700 according to the second embodiment classifies images in image units. It is an example of That is, the image data processing apparatus 1700 calculates human surrounding feature quantities indicating features of pixels around a person shown in an image, and based on the calculated human surrounding feature quantities, one image is classified into different classification destinations. Classify into any of the classified events of the event.
- ⁇ Configuration> ⁇ Hardware Configuration of Image Data Processing Device 1700>
- the hardware configuration of the image data processing apparatus 1700 is the same as the configuration of the image data processing apparatus 100 according to the first embodiment. Therefore, the description is omitted.
- FIG. 17 is a functional block diagram showing a configuration of main functional blocks of the image data processing apparatus 1700.
- the image data processing apparatus 1700 accepts the family scene information write / read unit 204, the family scene information calculation unit 206, and the event name information reception.
- Unit 211 and family scene information storage unit 233 are deleted, image group data reception unit 201 is changed to image data reception unit 1701, image group classification unit 208 is changed to image classification unit 1708, and image storage unit 231 is It is changed to the image storage unit 1731 and the event feature information storage unit 234 is changed to the event feature information storage unit 1734.
- the image data accepting unit 1701 is a part of the function of the image group data accepting unit 201 according to the first embodiment, and is realized by the CPU 101 executing a program. And a function of reading a designated image from the user, and a function of giving an image ID for specifying the image when the image is read.
- an image may be read from an external communication device via 180.
- the image classification unit 1708 is realized by the CPU 101 executing a program, and a part of the functions of the image group classification unit 208 according to the first embodiment is modified.
- the image storage unit 1731 is obtained by changing part of the directory structure of the image storage unit 231 according to the first embodiment, and is connected to the image writing / reading unit 202.
- FIG. 18 is a directory structure diagram showing the directory structure of the image storage unit 1731. As shown in FIG.
- the directory structure of the image storage unit 1731 comprises a total of two layers, the highest hierarchy 1810 and the first directory hierarchy 1820.
- the first directory hierarchy 320 there are a plurality of classification destination event directories such as a ski directory 1821, a swimming directory 1822 and a picnic directory 1823, and an actual data storage directory 1824.
- the actual data storage directory 324 is a directory for storing image data, and the data of the image is stored only in the actual data storage directory 324.
- the classification destination event directory is a directory having the same name as the classification destination event to which the image is classified, and there is only one directory of the same name.
- each classification destination event directory the data indicating the image is linked by holding information indicating the address of the image data classified into the classification destination event having the same name as the classification destination event directory. Directory.
- the event feature information storage unit 1734 is a part of the event feature information storage unit 234 according to the first embodiment in which a part of the event feature information stored therein is changed, and is connected to the event feature information write / read unit 209. Be done.
- FIG. 19 is a view showing an example of the data structure of event feature information stored in the event feature information storage unit 1734. As shown in FIG.
- the event feature information is image group family scene black ratio 0.5 or more 1901, image group family scene blue ratio 0.5 or more 1902, image group family scene ratio green 0.5 or more 1903, image
- classification conditions 1900 such as group family scene ratio white 0.5 or more 1904 is associated with each of classification destination events 1910 such as fireworks 1911, sea bath 1912, picnic 1913, skis 1914 and the like.
- Characteristic operations performed by the image data processing apparatus 1700 include deformation image feature information generation processing and image classification processing.
- the image feature information generation process is a process in which the image data processing device 1700 reads an image and generates image feature information for the read image.
- FIG. 20 is a flowchart of deformed image feature information generation processing performed by the image data processing apparatus 1700.
- the deformed image feature information generation process is started when the remote control 197 receives an operation from the user to start the deformed image feature information generation process.
- the image data receiving unit 1701 starts reading an image (step S2000).
- the image data reception unit 1701 is connected to an external recording medium mounted on the external recording medium reading / writing device 140, or an external device via an USB cable 195 connected to the USB control device 150, or to the network 194. Images can be read from the communication device 180 that has been
- the image data reception unit 1701 reads an image recorded in the SD memory card 191, assigns a unique image ID to the read image, associates the image data with the image ID, and writes the image. It is written in the actual data storage directory 1824 of the image storage unit 1731 using the reading unit 202.
- face extraction unit 205 reads the image accepted by image group data acceptance unit 201 from image storage unit 231 using image write / read unit 202, and the image read out Decode in JPEG format.
- the face extraction unit 205 tries to recognize the face included in one image by referring to the face model to be held in the read image (step S2010).
- the face extraction unit 205 calculates face feature amounts for each of the recognized faces (step S2020). That is, the area of the recognized face area and the position of the recognized face are calculated, and the face ID for identifying the recognized face is sequentially assigned to each recognized face, and the recognized face is identified.
- the feature and the feature of the face included in the sample image stored in the sample image storage unit 236 are extracted, and the sample image stored in the sample image storage unit 236 is the same as the recognized feature of the face
- the corresponding face ID corresponding to the sample image is assigned to the corresponding face ID of the recognized face. If there is no image having the same facial feature as the recognized facial feature in the sample image stored in the sample image storage unit 236, it is indicated that the corresponding face ID of the recognized face is another person Grant "99" which means
- the face extraction unit 205 calculates a body region below the recognized face region for each of the recognized faces, and calculates a body feature value for each of the calculated body regions. (Step S2030). That is, the area of the calculated body area and the position of the calculated body area are calculated, and body IDs for specifying the calculated body area are sequentially assigned to each of the calculated body areas. Do.
- the human surrounding feature quantity extraction unit 207 calculates a face surrounding area based on the face area, calculates a body surrounding area based on the body area, and the calculated face surrounding area.
- the human-peripheral area is calculated from the calculated per-body area (step S2040).
- the human surrounding feature quantity extraction unit 207 calculates a human surrounding feature quantity based on the pixel values of the pixels included in the human surrounding area (step S2050). That is, for each pixel included in the human surrounding area, the color of the pixel is specified from each luminance of R, G, and B which are color components constituting the pixel, and for each specified color, the human surrounding area The ratio of the number of pixels specified for that color to the total number of pixels contained in is calculated as the color ratio of that color.
- step S2010 when the face extraction unit 205 does not recognize the face (step S2010: No), the face extraction unit 205 sets a null value as the value of each component of the face feature amount, and the body feature A null value is set as the value of each component of the amount, and the human surrounding feature quantity extraction unit 207 sets a null value as the value of the face surrounding area, and a null value as the value of the body surrounding area. A null value is set as the value of each component of the quantity (step S2060).
- the human surrounding feature quantity extraction unit 207 When the process of step S2050 is completed, or when the process of step S2060 is completed, the human surrounding feature quantity extraction unit 207 generates image feature information for the target image, and generates the generated image feature information.
- the image feature information storage unit 1732 stores the image feature information write / read unit 203 (step S2070), and the image data processing apparatus 1700 ends the deformed image feature information generation process.
- the image classification process is a process in which the image data processing device 1700 classifies an image into any of the classification destination events.
- FIG. 21 is a flowchart of image classification processing performed by the image data processing apparatus 1700.
- the image group classification process is started when the image data processing apparatus 1700 ends the deformed image feature information generation process.
- the image classification unit 1708 uses the image feature information writing and reading unit 203 to send the image feature of the image targeted by the deformed image feature information generation process from the image feature information storage unit 232.
- the information is read out, and event characteristic information is read out from the event characteristic information storage unit 1734 using the event characteristic information write / read unit 209 (step S2100).
- the image classification unit 1708 compares the read image feature information with the read event feature information (step S2110), and calculates the classification destination event of the image (step S2110). Step S2120). That is, it is checked whether the color ratio (see FIG. 6) included in the image feature information matches the classification condition 1900 (see FIG. 19) included in the event feature information, and the event feature information is included. If a match is found with the classification condition 1900, the classification destination event 1910 corresponding to the classification condition 1900 matching is calculated as the classification destination event to which the image is to be classified, and is included in the event feature information. If no match to the classification condition 1900 is found, a classification destination event called another event is calculated as a classification destination event to which the image group is to be classified.
- a classification destination event called another event is calculated as a classification destination event to which the image is to be classified.
- all matching destination events 1910 are calculated as the destination events to which the image is to be assigned.
- the image classification unit 1708 using the image writing / reading unit 202, is a target under the classification destination event directory corresponding to the classification destination event to be classified in the image storage unit 1731.
- the image is classified by holding information indicating the address of the data of the image and setting the data of the image to be linked (step S2130).
- the classification result output unit 210 causes the display 193 to display the classification destination event name of the classification destination event to be classified calculated by the image group classification unit 208 together with the event name associated with the image group.
- the data processing apparatus 100 ends the image group classification process.
- digital photographs can be stored as data. May be encoded using, for example, the PNG (Portable Network Graphics) method, the GIF (Graphics Interchange Format) method, etc., or may be bitmap data which is not encoded. .
- PNG Portable Network Graphics
- GIF Graphics Interchange Format
- a digital photograph is shown as an example as the content, as long as the image can be stored as digital data, it may be, for example, data of a picture read by a scanner.
- the interface 108, the communication device interface 109, the decoder 111, and the bus line 120 are integrated in the system LSI 110, if the same function as the system LSI 110 can be realized, they are necessarily integrated in one LSI.
- the decoder 111 is a DSP, it does not have to be a DSP if it has a function of decoding encoded data.
- the CPU 101 is shared.
- the CPU 101 may be a CPU different from the CPU 101, or may be a dedicated circuit configured with an ASIC or the like.
- the input device 170 has the function of receiving the operation command from the user wirelessly transmitted from the remote control 197, but if there is a function of receiving the operation command from the user Even if the configuration does not necessarily have the function of receiving the operation command wirelessly transmitted from the remote control 197, for example, it has the keyboard and the mouse, and has the function of receiving the operation command from the user via the keyboard and the mouse. It may be a configuration including a button group and a function of receiving an operation command from the user via the button group.
- the image group data accepting unit 201 accepts designation of two or more images from the user, and the designated image group is regarded as an image group included in one image group.
- the image group data receiving unit 201 receives the image data and the list of the images belonging to the image group, and based on the received list, the image and The configuration may be such that the image group is associated with the image group.
- the image group data accepting unit 201 sequentially assigns the image ID to the read image. However, if the image ID can be assigned without duplication, the image ID is not necessarily sequentially assigned. You do not need to give it.
- the model of the face is, for example, the luminance of parts forming the face such as eyes, nose and mouth, and information on relative positional relationship, etc.
- feature values may be used, for example, feature values calculated using a Gabor filter, and facial features such as eye color, position of mole, skin color, etc. Or a combination of those representing a plurality of facial features.
- black, blue, green and white have been exemplified as the main colors specified by the human surrounding feature quantity extraction unit 207, but it is not necessary to be limited to these colors. It does not matter.
- RGB, L * a * b, etc. may be used as the color space.
- the human-surrounding feature amount is configured based on the color included in the image, but if it indicates the feature of the image, the color included in the image is not necessarily included.
- the human surrounding feature quantity extraction unit 207 calculates the human surrounding feature quantity based on each pixel included in the human surrounding area. So that the pixel values of the pixels included in the fixed region of are more reflected in the human surrounding feature amount than the pixel values of the pixels not included in the fixed region around the face region. As long as the calculation can be performed, it is not necessarily limited to the case of calculating the human-surrounding feature amount based on each pixel included in the human-surrounding region.
- the human surrounding feature quantity extraction unit 207 causes, for all the pixels included in the image, to be weighted more as the distance from the face area is shorter according to the distance from the face area.
- the calculation may be performed based on weighted pixel values.
- the human-peripheral area is an area obtained by excluding the face area and the body area from the area included in at least one of the face-peripheral area and the body-peripheral area, If it is a fixed area around the area, it is not necessary to necessarily exclude the area of the face and the area of the body from the area included in at least one of the area around the face and the area around the body; It may be an area excluding the face area from the area, or it may be the body area itself, or an area excluding the face area from the body peripheral area and the body area.
- the shape of the human-surrounding area is not necessarily limited to a rectangle, and may be, for example, a hexagon or a circle.
- the face surrounding area is obtained by adding the horizontal width in the image of the face area to the left and the right in the horizontal direction in the image with respect to the area of the face.
- the area of the face is a rectangular area obtained by adding the width in the vertical direction of the image to the upper side, if it is a fixed area around the area of the face, it is not always horizontal in the image with respect to the face area.
- the horizontal width of the face area in the image is added to the left and right of the direction respectively, and it is limited to the rectangular area of the face in the vertical direction above the image plus the vertical width in the image There is no need, for example, adding half of the horizontal width in the image of the face region to the left and right of the horizontal direction in the image with respect to the face region, and the face in the vertical direction of the image Area of It may be a rectangular region, such as a plus half the width of the vertical direction in the image, and further, may be a region having a shape other than a rectangle.
- the area of the recognized face is a rectangular area with the smallest area among the rectangles including the recognized face and the sides in the horizontal direction in the image and the sides in the vertical direction in the image.
- the rectangular area having the smallest area among the rectangles having the horizontal side in the image and the vertical side in the image is necessarily described. For example, it may be a region surrounded by a curve along the contour of the face.
- the width of the body area is 1.5 times the horizontal width of the image of the face area below the recognized face area
- the vertical width of the image of the face area is Although it is assumed that the area is a rectangular area twice as large as the area, if it is an area estimated to have a body, the horizontal width in the image of the face area below the recognized face area is necessarily 1.5. It is not necessary to be limited to a rectangular area in which the vertical width in the image of the face area is doubled, for example, the horizontal width in the image of the face area below the recognized face area. It may be a rectangular area which is doubled and which is 1.5 times the width in the vertical direction in the image of the face area.
- the body is detected by the image recognition process and indicated by the recognized body.
- the body peripheral area is obtained by adding the horizontal width in the image of the face area to the left and right in the horizontal direction in the image with respect to the body area, and although it is assumed that each of the upper and lower areas is a rectangular area obtained by adding the width of half of the width in the vertical direction of the image to the area of the face, it is not limited to the area of the body if it is a fixed area around the area of the body.
- the horizontal width in the image of the face area is added to each of the left and right in the horizontal direction in the image, and half of the vertical width in the image in the image It is not necessary to be limited to a rectangular area added with the width of, for example, the width of half of the horizontal width in the image of the area of the body on the left and right of the horizontal direction in the image with respect to the area of the body.
- the area of the body to may be a rectangular region obtained by adding the width of the vertical direction in the image, and further, may be a region having a shape other than a rectangle.
- the image family scene feature quantity is calculated by dividing the value of each color ratio of the human surrounding feature quantity by the value of the face area, but the size of the area of the person in the image is large.
- the value of each color ratio of the human surrounding feature is not necessarily the face area It is not necessary to calculate by dividing by the value of, for example, it may be calculated by dividing the value of each color ratio of the feature amount of human surroundings by the sum of the value of the face area and the value of the body area. Absent. (17) In the first embodiment, the image group family scene feature is the average value of the image family scene feature corresponding to the image including the face recognized as the face of the family.
- the image is calculated without including the image family scene feature of the image not included, it does not necessarily have to be the average value of the image family scene feature corresponding to the image including the face recognized as the face of the family
- it may be an average value of image family scene feature amounts of all images including a recognized face, or an average value of image family scene feature amounts of an image including a recognized face indicating a specific person. It does not matter.
- the image group family scene feature may be calculated not as the average value of the image family scene feature in the corresponding image group but as weighted according to the recognized face included in the image. .
- the image group classification unit 208 classifies the image groups based on the image group family scene information and the event feature information, but at least the image groups are classified based on the image group family information. For example, image groups need not necessarily be classified based on event feature information.
- image group family scene information may be learned as a teacher, and image groups may be classified based on the learning results.
- the learning method can be realized, for example, by a method performed using a learning model such as a logistic regression analysis method or a SVM (Support Vector Machine) method. (19)
- a learning model such as a logistic regression analysis method or a SVM (Support Vector Machine) method.
- the human-surrounding feature value extraction unit 207 calculates one human-surrounding area when a plurality of faces recognized in one image is included, but at least one human-surrounding area is calculated. If the area around people is calculated, the area around people to be calculated does not necessarily have to be limited to one.
- human surrounding regions are calculated for each of the recognized faces, and human peripheral feature amounts are calculated for each of the human surrounding regions.
- various methods can be considered as a method of calculating an image family scene feature quantity of the image.
- an image family scene feature amount corresponding to the person (hereinafter referred to as "periphery image family scene feature amount”) is calculated and calculated
- a method of using an average value as the image family scene feature value of the image a method of calculating the image family scene feature value only from the person surrounding information of the person indicated by the specific face ID, a specific position (for example, the center of the screen, screen Image family scene feature amount by weighting each person's surrounding image family scene feature amount according to a method of calculating the image family scene feature amount from the person ambient information of the person at the right end, etc.
- a method of calculating etc. can be considered.
- the sample image storage unit 236 stores image data of an image including a specific person, but the face extraction unit 205 may extract the feature of the face of the specific person. If what can be stored is stored, it is not necessary to store an image including a specific person. For example, the face feature of the specific person may be stored.
- the image group classification unit 208 determines the classification destination event to be classified, it is associated with the image group under the event directory corresponding to the classification destination event to be classified. An image directory is classified by creating an event directory having the same name as an event name and linking data of all the images belonging to the image group under the event directory.
- the image data processing device 100 classifies the image group based on the image group family scene information and the event feature information, but at least the image group family scene information and the event feature As long as it can be performed based on the information, it does not necessarily have to be performed based only on the image group family scene information and the event feature information.
- the image data processing apparatus 100 further has a function of calculating image group scene feature information indicating the feature of the image belonging to the image group based on the feature amount of the entire image, and this image group scene feature information
- the image group may be classified based on the image group family scene information and the event feature information.
- the image data processing device 100 is further based on the image group family scene information and the event feature information.
- the second classification may be performed in more detail.
- the image group sorting unit 208 sets all the matching destination events 910 as matching destination events for which the image group should be sorted. Although it is calculated, if it is possible to calculate at least one classification destination event into which the image group is to be classified, all of the matching classification destination events 910 are necessarily calculated as the classification destination event into which the image group is to be classified There is no need.
- the image group family scene color ratio having the largest value among the image group family scene color ratios matching the classification condition 900 is an example of a method of calculating the classification destination event in which the image group is to be classified.
- a conceivable method is to calculate a classification destination event corresponding to a classification condition that matches the above as a classification destination event to which an image group is to be classified.
- the image data processing apparatus is a server apparatus that provides a network service.
- the server apparatus receives content from the AV apparatus, personal computer, digital camera or the like in which the content is stored via the network, the image data processing according to the method described in the above embodiment for the received content.
- the processing result may be transmitted to an AV device, a personal computer, a digital camera or the like via a network.
- the transmission destination of the processing result may be for the device that received the content or for other devices. Specifically, another device owned by the user of the device receiving the content, a device owned by a family or friend of the user of the device receiving the content, a server device providing network services such as SNS or image sharing service, etc. Can be mentioned.
- the processing result may be stored in the server apparatus itself which provides the method described in the above embodiment as a network service.
- a control program comprising program code for causing the CPU of the image data processing apparatus and various circuits connected to the CPU to execute the image group classification operation and the like described in the first and second embodiments.
- the information may be recorded on a recording medium, or may be distributed via various communication paths.
- Such recording media include an IC card, a hard disk, an optical disk, a flexible disk, a ROM, and the like.
- the control program distributed and distributed is used by being stored in a memory or the like that can be read by the CPU, and the CPU executes the control program to realize various functions as shown in each embodiment.
- part of the control program is transmitted to a device (CPU) that can execute a program separate from the image classification device via various communication paths etc., and part of the control program in the device that can execute the program. May be performed.
- An image classification device is an image data processing device that calculates image feature information for classifying an image, and is a face identification unit that identifies a face area included in one image. And an image feature calculation unit that calculates image feature information in the image from an image feature amount calculated based on at least a part of pixels of one image, and the image feature calculation unit The image feature quantity calculated based on the pixels included in the fixed area around the identified face area is more than the image feature quantity calculated based on the pixels not included in the fixed area The image feature information is calculated so as to be greatly reflected in the feature information.
- a photographer of an image shoots an image including a human face at an event, it tends to shoot so that the feature of the event appears in an area around the face of the person. For example, in the sea bathing, the photographer tends to shoot an image so that the area around the person's face is blue-blue in the area around the person's face. In ski travel, the area around the person's face has snow white. There is a tendency for the image to be taken to be more.
- the image data processing apparatus can set the image feature quantity calculated based on the pixels of the area around the face of the person who tends to easily show the feature of the event.
- Image feature information is extracted by placing more importance on the image feature amount calculated based on the pixels of an area away from the face of a person who tends not to show features.
- the image data processing apparatus can calculate image feature information in which the feature of the event is more reflected than the conventional image data processing apparatus.
- this image data processing apparatus can improve the classification accuracy in the case of classifying an image using the image feature information calculated from the image data processing apparatus, as compared with the conventional case.
- FIG. 22 is a functional block diagram showing a functional configuration of the image data processing device 2200 in the modification.
- the image data processing apparatus 2200 is for calculating image feature information for classifying an image, and as shown in FIG. 22, it comprises a face identification unit 2201 and an image feature calculation unit 2202. .
- the face identification unit 2201 is connected to the image feature calculation unit 2202 and has a function of identifying a face area included in one image. As an example, it is realized as face extraction unit 205 in the first embodiment.
- the image feature calculation unit 2202 is connected to the face identification unit 2201 and has a function of calculating image feature information in the image from an image feature amount calculated based on at least a part of pixels of one image. Then, the image feature calculation unit 2202 determines that the image feature amount calculated based on the pixels included in the fixed area around the area of the face specified by the face specifying unit 2201 is not included in the fixed area.
- the image feature information is calculated so as to be reflected to the image feature information more largely than the image feature amount calculated based on pixels.
- it is realized as the human surrounding feature quantity extraction unit 207 in the first embodiment.
- the image feature calculation unit calculates the image feature information based on pixels other than the pixels included in the area of the face identified by the face identification unit and the pixels not included in the predetermined area. It may be performed from the image feature amount calculated by
- the image feature information calculated based on the pixels included in the face area and the image feature calculated based on the pixels not included in the fixed area are image feature information. It can be made to not be reflected.
- the face specifying unit specifies a rectangular area having a side in the horizontal direction and a side in the vertical direction in the image including the face as the area of the face, and the image feature calculation unit
- the area of the face identified by the face identification unit is enlarged at least in the upper direction of the image, the rightward direction in the image, and the left direction in the image by a width determined by a predetermined algorithm.
- the image feature information may be calculated with the rectangular extended face area as the fixed area.
- This configuration makes it possible to specify the area of the face using the coordinates of the two corners that are diagonal in the rectangle, and also makes it possible to use a rectangular area around the face as a rectangle. It becomes possible to specify using the coordinates of two corners that are diagonal at.
- an image classification device is an image data processing device that calculates image feature information for classifying an image, and a face that specifies a face area included in one image.
- An image feature calculation unit that calculates image feature information in the image from an image feature amount calculated based on at least a part of pixels of an image, and an area of the face identified by the face identification unit
- a body specifying unit for specifying an area defined by the predetermined algorithm to the area of the face as the body area below, the image feature calculating unit is included in the body region specified by the body specifying unit
- the image feature information is calculated so that the image feature amount calculated based on pixels is more greatly reflected in the image feature information than the image feature amount calculated based on pixels not included in the body region. of And performing out.
- the feature tends to appear so that the feature of the event appears in an area around the person's body.
- the photographer tends to take images so that the area around the person's body is bluer in the sea, and in ski travel, the area around the person's body has snow white.
- images to be taken are more.
- the image data processing apparatus having the above-described configuration is characterized in that the image feature quantity calculated based on the pixels of the area around the body of the person who tends to show the feature of the event Image feature information is extracted by placing more importance on the image feature amount calculated based on the pixels of an area away from the body of a person who tends not to show features.
- the image data processing apparatus can calculate image feature information in which the feature of the event is more reflected than the conventional image data processing apparatus.
- this image data processing apparatus can improve the classification accuracy in the case of classifying an image using the image feature information calculated from the image data processing apparatus, as compared with the conventional case.
- FIG. 23 is a block diagram showing a functional configuration of the image data processing device 2300 in the above modification.
- the image data processing apparatus 2300 is for calculating image feature information for classifying an image, and as shown in FIG. 23, a face identification unit 2301, a body identification unit 2302, and an image feature calculation unit 2303. And consists of
- the face identification unit 2301 is connected to the body identification unit 2302 and has a function of identifying a face area included in one image. As an example, it is realized as a part that realizes function 1 and function 2 in face extraction section 205 in the first embodiment.
- the body specifying unit 2302 is connected to the face specifying unit 2301 and the image feature calculating unit 2303, and is an area defined by a predetermined algorithm for the area of the face under the area of the face specified by the face specifying unit 2301. It has a function to identify as a body region. As an example, it is realized as a part that realizes function 3 in face extraction section 205 in the first embodiment.
- the image feature calculation unit 2303 is connected to the body identification unit 2302, and has a function of calculating image feature information in the image from the image feature amount calculated based on at least a part of pixels of one image. Then, the image feature calculation unit 2303 calculates the image feature amount calculated based on the pixels included in the body region identified by the body identification unit 2302 based on the pixels not included in the body region.
- the image feature information is calculated so as to be reflected to the image feature information more largely than the image feature amount. As an example, it is realized as a portion for realizing the functions 4 and 5 in the human surrounding feature quantity extraction unit 207 in the first embodiment.
- the image feature calculation unit may calculate the image feature information from an image feature amount calculated based on pixels included only in the body region identified by the face identification unit. .
- the image feature calculation unit may perform at least the upward direction in the image, the rightward direction in the image, and the leftward direction in the image with respect to the body region identified by the body identification unit.
- the image feature information is calculated by including, in the body region, an extended body region in which the face region identified by the face identification unit is excluded from the region expanded by a width determined by a predetermined algorithm It may be
- Such a configuration allows the body region to include an area around the body of the person corresponding to the body region.
- the image feature calculation unit calculates the image feature information when the face identification unit identifies the first face area and the second face area in one image.
- a pixel not included in at least one of the constant region corresponding to the first face region and the constant region corresponding to the second face region, and a pixel included in the first face region It may be performed from the pixel values of pixels other than the pixels included in the second face area.
- the image feature calculating unit may calculate the image feature information as the image feature information. It may be performed from the image feature amount calculated based on the pixels included in the first body region and the second body region.
- the pixel values of the pixels included in the first body region and the second body region are used. Both pixel values of the included pixel and the pixel value of the included pixel can be reflected in the image feature information.
- an image group feature calculation unit for calculating feature information wherein the face identification unit further calculates face area information indicating an area ratio which is a ratio of an area of a face area to be identified to an area of an image;
- the image group feature calculation unit is based on the first image whose face area information has been calculated by the face identification unit and the area ratio of the face area indicated by the face area information of the first image by the face identification unit.
- the image feature information of the first image is more than the image feature information of the second image. As reflected largely to information, it may perform the calculation of the image characteristic information.
- image group feature information for classifying images in image group units is reflected to a larger extent as image feature information of an image having a smaller area of a face area included in the image. To be able to calculate.
- the image group classification unit classifies an image group into any of a plurality of classification destination categories, and the image group classification unit calculates image group feature information according to the image group feature information. Classification of the image group with respect to the image group, reference information indicating the range of the feature indicated by the image group feature information in each of the plurality of classification destination categories, and the image group feature calculated by the image group feature calculation unit It may be performed based on the information.
- images can be classified in image group units.
- the image feature calculation unit includes information related to a color included in the image in the image feature information to be calculated, and the image group feature calculation unit calculates the image group feature information to be calculated.
- Information on a color included in the image to which it belongs is included, and the previous reference information includes color reference information indicating the range of the feature of the color indicated by the image group feature information in each of the plurality of classification destination categories;
- the unit calculates the classification of the image group to be performed on the image group for which the image group characteristic information is calculated by the image group characteristic information, by color reference information included in the reference information, and the image group characteristic calculation unit. It may be performed based on the information related to the color included in the image group feature information.
- images can be classified in image group units based on information on colors.
- the image data processing apparatus can be widely applied to devices having a function of storing a plurality of digital images.
- Image Data Processing Device 201 Image Group Data Reception Unit 202 Image Writing / Reading Unit 203 Image Feature Information Writing / Reading Unit 204 Family Scene Information Writing / Reading Unit 205 Face Extraction Unit 206 Family Scene Information Calculation Unit 207 Human Surrounding Feature Amount Extraction Unit 208 image group classification unit 209 event feature information write / read unit 210 classification result output unit 211 event name information reception unit 212 event feature information reception unit 213 sample image reception unit 214 sample image writing unit 231 image storage unit 232 image characteristic information storage Unit 233 Family scene information storage unit 234 Event feature information storage unit 236 Sample image storage unit
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
Description
以下、本発明に係る画像データ処理装置の一実施形態として、画像に写る人物の周囲の画素の特徴を示す人周囲特徴量を算出し、算出した人周囲特徴量に基づいて、一つの行事であるイベントに関連して撮影された複数枚の画像からなる画像グループ単位で、画像を互いに異なる予め定められた複数の分類先イベントのうちのいずれかの分類先イベントに分類する画像データ処理装置100について説明する。
<画像データ処理装置100のハードウエア構成>
図1は、画像データ処理装置100の主要なハードウエア構成を示すハードウエアブロック図である。
から符号化された画像データを読み出して復号し、復号した画像データをディスプレイ193に出力する機能等を実現する。
図2は、画像データ処理装置100の主要な機能ブロックの構成を示す機能ブロック図である。
画像データ処理装置100の行う特徴的な動作に、画像特徴情報生成処理と、画像家族シーン情報生成処理と、画像グループ家族シーン情報生成処理と、画像グループ分類処理とがある。
画像特徴情報生成処理は、画像データ処理装置100が、画像グループ単位で画像を読み込み、読み込んだ画像のそれぞれに対して、画像特徴情報を生成する処理である。
画像家族シーン情報生成処理は、画像データ処理装置100が、画像特徴情報に基づいて、画像家族シーン情報を生成する処理である。
画像グループ家族シーン情報生成処理は、画像データ処理装置100が、画像グループに属する各画像の画像家族シーン情報に基づいて、その画像グループの画像グループ家族シーン情報を生成する処理である。
画像グループ分類処理は、画像データ処理装置100が、画像グループを、分類先イベントのいずれかに分類する処理である。
以下、具体例を用いて、画像データ処理装置100の特徴について補足説明を行う。
図16は、2枚の画像を示す図である。
<実施の形態2>
以下、本発明に係る画像データ処理装置の一実施形態として、実施の形態1に係る画像データ処理装置100の一部を変形した画像データ処理装置1700について図17を用いて説明する。
<画像データ処理装置1700のハードウエア構成>
画像データ処理装置1700のハードウエア構成は、実施の形態1に係る画像データ処理装置100の構成と同一のものである。よって、説明を省略する。
図17は、画像データ処理装置1700の主要な機能ブロックの構成を示す機能ブロック図である。
画像データ処理装置1700の行う特徴的な動作に、変形画像特徴情報生成処理と、画像分類処理とがある。
画像特徴情報生成処理は、画像データ処理装置1700が、画像を読み込み、読み込んだ画像に対して、画像特徴情報を生成する処理である。
画像分類処理は、画像データ処理装置1700が、画像を、分類先イベントのいずれかに分類する処理である。
<補足>
以上、本発明に係る画像データ処理装置の一実施形態として、実施の形態1、実施の形態2において、画像の分類を行う画像データ処理装置の例について説明したが、以下のように変形することも可能であり、本発明は上述した実施の形態で示した通りの画像データ処理装置に限られないことはもちろんである。
(1)実施の形態1において、画像データ処理装置100が記憶する画像として、JPEG方式で符号化されたデータとしたが、デジタル写真をデータとして記憶することができるものであれば、JPEG方式以外の符号化方式、例えば、PNG(Portable Network Graphics)方式やGIF(Graphics Interchange Format)方式等で符号化されたものであっても構わないし、符号化されないビットマップ方式のデータであっても構わない。
(2)実施の形態1において、CPU101と、ROM102と、RAM103と、ハードディスク装置インターフェース104と、外部記録媒体読取書込装置インターフェース105と、USB制御装置インターフェース106と、出力装置インターフェース107と、入力装置インターフェース108と、通信装置インターフェース109と、デコーダ111と、バスライン120とが、システムLSI110に集積されているとしたが、システムLSI110と同じ機能を実現することができれば、必ずしも1つのLSIに統合されている必要はなく、例えば、複数の集積回路等で実現されていても構わない。
(3)実施の形態1において、デコーダ111は、DSPであるとしたが、符号化されたデータを復号する機能があれば、必ずしもDSPである必要はなく、例えば、CPU101が兼用する構成であっても構わないし、CPU101とは異なるCPUであっても構わないし、ASIC等で構成される専用回路であっても構わない。
(4)実施の形態1において、入力装置170は、リモコン197から無線で送信されるユーザからの操作コマンドを受け付ける機能を有する構成であるとしたが、ユーザからの操作コマンドを受け付ける機能があれば、必ずしもリモコン197から無線で送信される操作コマンドを受け付ける機能を有する構成でなくても、例えば、キーボードとマウスとを備え、キーボードとマウスとを介してユーザからの操作コマンドを受け付ける機能を有する構成であっても構わないし、ボタン群を備え、ボタン群を介してユーザからの操作コマンドを受け付ける機能を有する構成等であっても構わない。
(5)実施の形態1において、画像グループデータ受付部201が、ユーザからの、2枚以上の画像の指定を受け付け、指定された画像群を、1つの画像グループに含まれる画像群とするとしたが、画像と画像グループとの対応付けを取ることができれば、例えば、画像グループデータ受付部201は、画像データと、画像グループに属する画像のリストとを受け取り、受け取ったリストに基づいて、画像と画像グループとを対応付けるといった構成であっても構わない。
(6)実施の形態1において、画像グループデータ受付部201は、読み込んだ画像に対して、シーケンシャルに画像IDを付与するとしたが、重複を避けて付与することができれば、必ずしもシーケンシャルに画像IDを付与しなくても構わない。
(7)実施の形態1において、顔のモデルは、例えば、目、鼻、口等の顔を形成するパーツの輝度や、相対的な位置関係に関する情報等であるとしたが、顔を認識することができる情報であれば、これら以外、例えば、ガボールフィルタを用いて算出された特徴量を用いるものであっても構わないし、目の色や、ほくろの位置、肌の色等といった顔の特徴を示すものであっても構わないし、複数の顔の特徴を表すものの組み合わせであっても構わない。
(8)実施の形態1において、人周囲特徴量抽出部207が特定する主要色として、黒、青、緑、白を例示したが、これらの色に限られる必要はなく、例えば、赤、黄等であっても構わない。また、色空間としてRGBやL*a*b等を用いても良い。
(9)実施の形態1において、人周囲特徴量は、画像に含まれる色に基づいたもので構成されているとしたが、画像の特徴を示すものであれば、必ずしも、画像に含まれる色に基づいたもので構成されている必要はなく、例えば、輝度やテクスチャ特徴に基づいたもので構成されていても構わないし、写る物体に基づいたものであっても構わない。
(10)実施の形態1において、人周囲特徴量抽出部207は、人周囲特徴量を、人周囲領域に含まれる各画素に基づいて算出する場合の例について説明したが、顔の領域の周囲の一定領域に含まれる画素の画素値の方が、顔の領域の周囲の一定領域に含まれない画素の画素値よりも、人周囲特徴量へ大きく反映されるように、人周囲特徴量の算出を行うことができれば、必ずしも、人周囲特徴量を、人周囲領域に含まれる各画素に基づいて算出する場合に限られない。
(11)実施の形態1において、人周囲領域は、顔周囲領域と体周囲領域との少なくとも一方に含まれる領域から顔の領域と体の領域とを除外した領域であるとしたが、顔の領域の周囲の一定領域であれば、必ずしも、顔周囲領域と体周囲領域との少なくとも一方に含まれる領域から顔の領域と体の領域とを除外した領域である必要はなく、例えば、顔周囲領域から顔の領域を除外した領域であるとしても構わないし、体領域そのものであるとしても構わないし、体周囲領域から顔領域を除外した領域と体領域とからなる領域であるとしても構わない。
(12)実施の形態1において、顔周囲領域は、顔の領域に対して、画像における水平方向の左右のそれぞれに、顔の領域の、画像における水平方向の幅を加え、画像における垂直方向の上方に、顔の領域の、画像における垂直方向の幅を加えた矩形の領域であるとしたが、顔の領域の周囲の一定領域であれば、必ずしも、顔の領域に対して、画像における水平方向の左右のそれぞれに、顔の領域の、画像における水平方向の幅を加え、画像における垂直方向の上方に、顔の領域の、画像における垂直方向の幅を加えた矩形の領域に限定される必要はなく、例えば、顔の領域に対して、画像における水平方向の左右のそれぞれに、顔の領域の、画像における水平方向の幅の半分を加え、画像における垂直方向の上下のそれぞれに、顔の領域の、画像における垂直方向の幅の半分を加えた矩形の領域等であっても構わないし、さらには、矩形以外の形状の領域であっても構わない。
(13)実施の形態1において、認識した顔の領域は、認識した顔を含む、画像における水平方向の辺と画像における垂直方向の辺とを有する矩形のうち、面積が最小となる矩形の領域である場合の例について説明したが、認識した顔を含む領域であれば、必ずしも、画像における水平方向の辺と画像における垂直方向の辺とを有する矩形のうち、面積が最小となる矩形の領域に限定される必要はなく、例えば、顔の輪郭にそった曲線で囲まれた領域であっても構わない。
(14)実施の形態1において、体の領域は、認識した顔の領域の下方の、顔の領域の画像における水平方向の幅を1.5倍し、顔の領域の画像における垂直方向の幅を2倍した矩形の領域であるとしたが、体があると推定される領域であれば、必ずしも、認識した顔の領域の下方の、顔の領域の画像における水平方向の幅を1.5倍し、顔の領域の画像における垂直方向の幅を2倍した矩形の領域に限定される必要はなく、例えば、認識した顔の領域の下方の、顔の領域の画像における水平方向の幅を2倍し、顔の領域の画像における垂直方向の幅を1.5倍した矩形の領域であっても構わないし、さらには、画像認識処理により体を検出して、その認識された体によって示される領域であるとしても構わないし、さらには、矩形以外の形状の領域であっても構わない。
(15)実施の形態1において、体周囲領域は、体の領域に対して、画像における水平方向の左右のそれぞれに、顔の領域の、画像における水平方向の幅を加え、画像における垂直方向の上下のそれぞれに、顔の領域の、画像における垂直方向の幅の半分の幅を加えた矩形の領域であるとしたが、体の領域の周囲の一定領域であれば、必ずしも、体の領域に対して、画像における水平方向の左右のそれぞれに、顔の領域の、画像における水平方向の幅を加え、画像における垂直方向の上下のそれぞれに、顔の領域の、画像における垂直方向の幅の半分の幅を加えた矩形の領域に限定される必要はなく、例えば、体の領域に対して、画像における水平方向の左右のそれぞれに、体の領域の、画像における水平方向の幅の半分の幅を加え、画像における垂直方向の上下のそれぞれに、体の領域の、画像における垂直方向の幅を加えた矩形の領域であるとしても構わないし、さらには、矩形以外の形状の領域であっても構わない。
(16)実施の形態1において、画像家族シーン特徴量は、人周囲特徴量の各色比率の値を、顔面積の値で除算することで算出されるとしたが、画像における人物の面積の大きさがより小さな画像の方が、その画像の画像家族シーン特徴量の値がより大きな値となるように重み付けされることとなれば、必ずしも、人周囲特徴量の各色比率の値を、顔面積の値で除算することで算出される必要はなく、例えば、人周囲特徴量の各色比率の値を、顔面積の値と体面積の値との和で除算することで算出されるとしても構わない。
(17)実施の形態1において、画像グループ家族シーン特徴量は、家族の顔として認識された顔を含む画像に対応する画像家族シーン特徴量の平均値であるとしたが、認識された顔を含まない画像の画像家族シーン特徴量を含めないで算出されるものであれば、必ずしも、家族の顔として認識された顔を含む画像に対応する画像家族シーン特徴量の平均値である必要はなく、例えば、認識された顔を含む全ての画像の画像家族シーン特徴量の平均値であっても構わないし、特定の人物を示す認識された顔を含む画像の画像家族シーン特徴量の平均値であっても構わない。
(18)実施の形態1において、画像グループ分類部208は、画像グループ家族シーン情報とイベント特徴情報とに基づいて画像グループを分類するとしたが、少なくとも画像グループ家族情報に基づいて画像グループを分類すれば、必ずしも、イベント特徴情報に基づいて画像グループを分類する必要はなく、例えば、画像グループ家族シーン情報を教師として学習し、その学習結果に基づいて、画像グループを分類しても構わない。学習方法は、例えば、ロジスティック回帰分析法、SVM(Support Vector Machine)法等の学習モデルを用いて行う手法によって実現できる。
(19)実施の形態1において、認識された顔に対応する対応顔IDが家族を示す場合にその認識された顔の人物が家族であるとしたが、認識された顔が家族であると類推される場合にその認識された顔の人物が家族であるとすれば、必ずしも、認識された顔に対応する対応顔IDが家族を示す場合にその認識された顔の人物が家族であるとする必要はなく、例えば、その認識された顔の特徴と同じ顔の特徴を持つ顔を含む画像が、画像記憶部231に所定の枚数(例えば10枚)以上含まれている場合に、その認識された顔の人物が家族であるとしても構わない。
(20)実施の形態1において、人周囲特徴量抽出部207は、一枚の画像に認識された顔が複数含まれている場合に、1つの人周囲領域を算出するとしたが、少なくとも1つの人周囲領域を算出すれば、必ずしも算出する人周囲領域は1つに限定される必要はない。
(21)実施の形態1において、サンプル画像記憶部236は、特定の人物を含む画像の画像データを記憶するとしているが、顔抽出部205が、特定の人物の顔の特徴を抽出することができるものを記憶していれば、必ずしも、特定の人物を含む画像を記憶する必要はなく、例えば、特定の人物の顔の特徴そのものを記憶するとしても構わない。
(22)実施の形態1において、画像グループ分類部208は、分類されるべき分類先イベントを決定すると、分類されるべき分類先イベントに対応するイベントディレクトリの下に、画像グループに対応付けられているイベント名と同一名称のイベントディレクトリを作成して、そのイベントディレクトリの下に、画像グループに属する全ての画像のデータのリンクを張ることで、画像グループを分類するとしたが、画像グループに属する画像が同じ分類先イベントに対応付けられていれば、必ずしもリンクを張ることで画像グループを分類するとする必要はなく、例えば、画像グループに属する画像に、分類先イベントを特定するためのタグを付与するとしても構わない。
(23)実施の形態1において、画像データ処理装置100は、画像グループの分類を、画像グループ家族シーン情報とイベント特徴情報とに基づいて行うとしたが、少なくとも、画像グループ家族シーン情報とイベント特徴情報とに基づいて行うことができれば、必ずしも、画像グループ家族シーン情報とイベント特徴情報とだけに基づいて行う必要はない。
(24)実施の形態1において、画像グループ分類部208は、一致する分類先イベント910が複数ある場合には、一致する分類先イベント910の全てを、画像グループが分類されるべき分類先イベントとして算出するとしたが、画像グループが分類されるべき分類先イベントを少なくとも1つ算出することができれば、必ずしも、一致する分類先イベント910の全てを、画像グループが分類されるべき分類先イベントとして算出する必要はない。
(25)さらに、上記の実施の形態で説明した手法をネットワークサービスとして提供するサーバ装置とすることも可能である。この場合、画像データ処理装置を、ネットワークサービスを提供するサーバ装置とする。そして、このサーバ装置が、コンテンツが蓄積されたAV機器、パーソナルコンピュータ、デジタルカメラなどからネットワークを介してコンテンツを受信すると、受信したコンテンツに対して上記の実施の形態で説明した手法による画像データ処理を行い、その処理結果を、ネットワークを介してAV機器、パーソナルコンピュータ、デジタルカメラなどに送信するようにすればよい。なお、処理結果の送信先は、コンテンツを受信した機器に対してであってもそれ以外の機器に対してであってもよい。具体的には、コンテンツを受信した機器のユーザが所有する他の機器、コンテンツを受信した機器のユーザの家族や友人が所有する機器、SNSや画像共有サービスなどのネットワークサービスを提供するサーバ装置などが挙げられる。また、処理結果を送信する代わりに、あるいは処理結果を送信することに加えて、処理結果を、上記の実施の形態で説明した手法をネットワークサービスとして提供するサーバ装置自身に保存することとしてもよい。
(26)実施の形態1、実施の形態2で示した、画像グループ分類動作等を画像データ処理装置のCPU、及びそのCPUに接続された各種回路に実行させるためのプログラムコードからなる制御プログラムを、記録媒体に記録すること、又は各種通信路等を介して流通させ頒布することもできる。このような記録媒体には、ICカード、ハードディスク、光ディスク、フレキシブルディスク、ROM等がある。流通、頒布された制御プログラムはCPUに読み出され得るメモリ等に格納されることにより利用に供され、そのCPUがその制御プログラムを実行することにより各実施形態で示したような各種機能が実現されるようになる。なお、制御プログラムの一部を画像分類装置とは別個のプログラム実行可能な装置(CPU)に各種通信路等を介して送信して、その別個のプログラム実行可能な装置においてその制御プログラムの一部を実行させることとしてもよい。
(27)以下、さらに本発明の一実施形態に係る画像データ処理装置の構成及びその変形例と各効果について説明する。
201 画像グループデータ受付部
202 画像書込読出部
203 画像特徴情報書込読出部
204 家族シーン情報書込読出部
205 顔抽出部
206 家族シーン情報算出部
207 人周囲特徴量抽出部
208 画像グループ分類部
209 イベント特徴情報書込読出部
210 分類結果出力部
211 イベント名情報受付部
212 イベント特徴情報受付部
213 サンプル画像受付部
214 サンプル画像書込部
231 画像記憶部
232 画像特徴情報記憶部
233 家族シーン情報記憶部
234 イベント特徴情報記憶部
236 サンプル画像記憶部
Claims (14)
- 画像を分類するための画像特徴情報を算出する画像データ処理装置であって、
一画像に含まれる顔の領域を特定する顔特定部と、
一画像の少なくとも一部の画素に基づいて算出される画像特徴量から、当該画像における画像特徴情報を算出する画像特徴算出部とを備え、
前記画像特徴算出部は、前記顔特定部によって特定された顔の領域の周囲の一定領域に含まれる画素に基づいて算出される画像特徴量の方が、当該一定領域に含まれない画素に基づいて算出される画像特徴量よりも、前記画像特徴情報へ大きく反映されるように、
前記画像特徴情報の算出を行う
ことを特徴とする画像データ処理装置。 - 前記画像特徴算出部は、前記画像特徴情報の算出を、前記顔特定部によって特定された顔の領域に含まれる画素と前記一定領域に含まれない画素と以外の画素に基づいて算出された画像特徴量から行う
ことを特徴とする請求項1記載の画像データ処理装置。 - 前記顔特定部は、顔を含む、画像における水平方向の辺と画像における垂直方向の辺とを有する矩形の領域を、前記顔の領域として特定し、
前記画像特徴算出部は、前記顔特定部によって特定された顔の領域に対して、少なくとも、画像における上方向と画像における右方向と画像における左方向とのそれぞれの方向に、それぞれ所定のアルゴリズムで定められる幅だけ拡大された矩形の拡張顔領域を、前記一定領域として、前記画像特徴情報の算出を行う
ことを特徴とする請求項2記載の画像データ処理装置。 - 画像を分類するための画像特徴情報を算出する画像データ処理装置であって、
一画像に含まれる顔の領域を特定する顔特定部と、
一画像の少なくとも一部の画素に基づいて算出される画像特徴量から、当該画像における画像特徴情報を算出する画像特徴算出部と、
前記顔特定部によって特定された顔の領域の下方に、当該顔の領域に対して所定のアルゴリズムで定められる領域を体領域として特定する体特定部とを備え、
前記画像特徴算出部は、前記体特定部によって特定された体領域に含まれる画素に基づいて算出される画像特徴量の方が、当該体領域に含まれない画素に基づいて算出される画像特徴量よりも、前記画像特徴情報へ大きく反映されるように、前記画像特徴情報の算出を行う
ことを特徴とする画像データ処理装置。 - 前記画像特徴算出部は、前記画像特徴情報の算出を、前記顔特定部によって特定された前記体領域にのみ含まれる画素に基づいて算出された画像特徴量から行う
ことを特徴とする請求項4記載の画像データ処理装置。 - 前記画像特徴算出部は、前記体特定部によって特定された体領域に対して、少なくとも、画像における上方向と画像における右方向と画像における左方向とのそれぞれの方向に、それぞれ所定のアルゴリズムで定められる幅だけ拡大された領域から、前記顔特定部によって特定された顔の領域が除外された拡張体領域を、前記体領域に含ませて、前記画像特徴情報の算出を行う
ことを特徴とする請求項4記載の画像データ処理装置。 - 前記画像特徴算出部は、前記顔特定部が1つの画像に対して第1の顔の領域と第2の顔の領域とを特定した場合に、前記画像特徴情報の算出を、当該第1の顔の領域に対応する前記一定領域と当該第2の顔の領域に対応する前記一定領域との少なくとも一方に含まれない画素と、当該第1の顔の領域に含まれる画素と、当該第2の顔の領域に含まれる画素と以外の画素に基づいて算出される画像特徴量から行う
ことを特徴とする請求項1記載の画像データ処理装置。 - 前記画像特徴算出部は、前記体特定部が1つの画像に対して第1の体領域と第2の体領域とを特定した場合に、前記画像特徴情報の算出を、当該第1の体領域と当該第2の体領域に含まれる画素に基づいて算出される画像特徴量から行う
ことを特徴とする請求項4記載の画像データ処理装置。 - 一画像グループに属する画像の一部又は全部の2枚以上の画像についての、画像特徴算出部によって算出された画像特徴情報に基づいて、当該画像グループを分類するための画像グループ特徴情報を算出する画像グループ特徴算出部とを備え、
前記顔特定部は、さらに、画像の面積に対する、特定する顔の領域の面積の比率である面積率を示す顔面積情報を算出し、
前記画像グループ特徴算出部は、前記顔特定部によって顔面積情報を算出された第1の画像と、前記顔特定部によって、当該第1の画像の顔面積情報によって示される顔の領域の面積率よりも大きい面積率を示す顔面積情報を算出された第2の画像とについて、当該第1の画像の画像特徴情報の方が、当該第2の画像の画像特徴情報よりも、前記画像グループ特徴情報へ大きく反映されるように、前記画像特徴情報の算出を行う
ことを特徴とする請求項1記載の画像データ処理装置。 - 画像グループを、複数の分類先カテゴリのうちのいずれかに分類する画像グループ分類部を備え、
前記画像グループ分類部は、前記画像グループ特徴情報によって画像グループ特徴情報を算出された画像グループに対して、前記画像グループの分類を、前記複数の分類先カテゴリそれぞれにおける、画像グループ特徴情報の示す特徴の範囲を示す基準情報と、前記画像グループ特徴算出部によって算出された画像グループ特徴情報とに基づいて行う
ことを特徴とする請求項9記載の画像データ処理装置。 - 前記画像特徴算出部は、算出する画像特徴情報に、画像に含まれる色に係る情報を含ませ、
前記画像グループ特徴算出部は、算出する画像グループ特徴情報に、画像グループに属する画像に含まれる色に係る情報を含ませ、
前基準情報は、前記複数の分類先カテゴリそれぞれにおける、画像グループ特徴情報の示す色の特徴の範囲を示す色基準情報を含み、
前記画像グループ分類部は、前記画像グループ特徴情報によって画像グループ特徴情報を算出された画像グループに対して行う前記画像グループの分類を、前記基準情報に含まれる色基準情報と、前記画像グループ特徴算出部によって算出された画像グループ特徴情報に含まれる前記色に係る情報とに基づいて行う
ことを特徴とする請求項10記載の画像データ処理装置。 - 画像を分類するための画像特徴情報を算出する画像データ処理装置を用いて行う画像データ処理方法であって、
一画像に含まれる顔の領域を特定する顔特定ステップと、
一画像の一部又は全部の画素に基づいて算出される画像特徴量から、当該画像における画像特徴情報を算出する画像特徴算出ステップとを備え、
前記画像特徴算出ステップは、前記顔特定ステップによって特定された顔の領域の周囲の一定領域に含まれる画素に基づいて算出される画像特徴量の方が、当該一定領域に含まれない画素に基づいて算出される画像特徴量よりも、前記画像特徴情報へ大きく反映されるように、前記画像特徴情報の算出を行う
ことを特徴とする画像データ処理方法。 - コンピュータを、画像を分類するための画像特徴情報を算出する画像データ処理装置として機能させるための画像データ処理プログラムであって、
コンピュータを、
一画像に含まれる顔の領域を特定する顔特定部と、
一画像の一部又は全部の画素に基づいて算出される画像特徴量から、当該画像における画像特徴情報を算出する画像特徴算出部とを備え、
前記画像特徴算出部は、前記顔特定部によって特定された顔の領域の周囲の一定領域に含まれる画素に基づいて算出される画像特徴量の方が、当該一定領域に含まれない画素に基づいて算出される画像特徴量よりも、前記画像特徴情報へ大きく反映されるように、前記画像特徴情報の算出を行うことを特徴とする画像データ処理装置として機能させる
ことを特徴とする画像データ処理プログラム。 - 画像を分類するための画像特徴情報を算出する半導体集積回路であって、
一画像に含まれる顔の領域を特定する顔特定部と、
一画像の一部又は全部の画素に基づいて算出される画像特徴量から、当該画像における画像特徴情報を算出する画像特徴算出部とを備え、
前記画像特徴算出部は、前記顔特定部によって特定された顔の領域の周囲の一定領域に含まれる画素に基づいて算出される画像特徴量の方が、当該一定領域に含まれない画素に基づいて算出される画像特徴量よりも、前記画像特徴情報へ大きく反映されるように、前記画像特徴情報の算出を行う
ことを特徴とする半導体集積回路。
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2012554676A JP5890325B2 (ja) | 2011-01-28 | 2012-01-24 | 画像データ処理装置、方法、プログラム及び集積回路 |
| CN201280001015.9A CN102893304B (zh) | 2011-01-28 | 2012-01-24 | 图像数据处理装置、方法、程序及集成电路 |
| US13/578,392 US8737726B2 (en) | 2011-01-28 | 2012-01-24 | Image data processing device, method, program and integrated circuit |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2011-016483 | 2011-01-28 | ||
| JP2011016483 | 2011-01-28 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2012102015A1 true WO2012102015A1 (ja) | 2012-08-02 |
Family
ID=46580591
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2012/000410 Ceased WO2012102015A1 (ja) | 2011-01-28 | 2012-01-24 | 画像データ処理装置、方法、プログラム及び集積回路 |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US8737726B2 (ja) |
| JP (1) | JP5890325B2 (ja) |
| CN (1) | CN102893304B (ja) |
| WO (1) | WO2012102015A1 (ja) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8565482B2 (en) * | 2011-02-28 | 2013-10-22 | Seiko Epson Corporation | Local difference pattern based local background modeling for object detection |
| JP6212878B2 (ja) * | 2013-02-21 | 2017-10-18 | 株式会社リコー | 画像処理装置、画像処理システム及びプログラム |
| US9569440B2 (en) * | 2013-02-22 | 2017-02-14 | Sony Corporation | Method and apparatus for content manipulation |
| JP6387700B2 (ja) * | 2014-06-18 | 2018-09-12 | ソニー株式会社 | 情報処理装置、情報処理システム、情報処理装置の制御方法およびプログラム |
| US20160110381A1 (en) * | 2014-10-17 | 2016-04-21 | Fuji Xerox Co., Ltd. | Methods and systems for social media-based profiling of entity location by associating entities and venues with geo-tagged short electronic messages |
| TWI579718B (zh) * | 2016-06-15 | 2017-04-21 | 陳兆煒 | 圖形資源管理系統及方法與內儲圖形資源管理程式之電腦程式產品 |
| KR102787157B1 (ko) * | 2016-12-07 | 2025-03-26 | 삼성전자주식회사 | 셀프 구조 분석을 이용한 구조 잡음 감소 방법 및 장치 |
| JP7129383B2 (ja) * | 2019-07-03 | 2022-09-01 | 富士フイルム株式会社 | 画像処理装置,画像処理方法,画像処理プログラムおよびそのプログラムを格納した記録媒体 |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007317077A (ja) * | 2006-05-29 | 2007-12-06 | Fujifilm Corp | 画像分類装置および方法ならびにプログラム |
| JP2010273328A (ja) * | 2009-04-20 | 2010-12-02 | Fujifilm Corp | 画像処理装置、画像処理方法およびプログラム |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH04232774A (ja) | 1990-12-28 | 1992-08-21 | Mitsubishi Paper Mills Ltd | 改ざん防止用感圧記録シート |
| JPH04315344A (ja) | 1991-04-15 | 1992-11-06 | Nippon Telegr & Teleph Corp <Ntt> | 符号誤り監視装置 |
| US7508961B2 (en) * | 2003-03-12 | 2009-03-24 | Eastman Kodak Company | Method and system for face detection in digital images |
| JP4315344B2 (ja) | 2003-11-27 | 2009-08-19 | 富士フイルム株式会社 | 画像編集装置および方法並びにプログラム |
| JP2006295890A (ja) * | 2005-03-15 | 2006-10-26 | Fuji Photo Film Co Ltd | アルバム作成装置、アルバム作成方法、及びプログラム |
| JP4232774B2 (ja) | 2005-11-02 | 2009-03-04 | ソニー株式会社 | 情報処理装置および方法、並びにプログラム |
| JP4466585B2 (ja) * | 2006-02-21 | 2010-05-26 | セイコーエプソン株式会社 | 対象物が表現された画像の数の算出 |
| CN101334780A (zh) * | 2007-06-25 | 2008-12-31 | 英特维数位科技股份有限公司 | 人物影像的搜寻方法、系统及存储影像元数据的记录媒体 |
| JP2010250634A (ja) * | 2009-04-17 | 2010-11-04 | Seiko Epson Corp | 画像サーバー、画像検索システム、画像検索方法およびインデックス作成方法 |
| WO2012017620A1 (ja) * | 2010-08-04 | 2012-02-09 | パナソニック株式会社 | 画像分類装置、方法、プログラム、プログラムを記録する記録媒体及び集積回路 |
-
2012
- 2012-01-24 WO PCT/JP2012/000410 patent/WO2012102015A1/ja not_active Ceased
- 2012-01-24 JP JP2012554676A patent/JP5890325B2/ja not_active Expired - Fee Related
- 2012-01-24 CN CN201280001015.9A patent/CN102893304B/zh not_active Expired - Fee Related
- 2012-01-24 US US13/578,392 patent/US8737726B2/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007317077A (ja) * | 2006-05-29 | 2007-12-06 | Fujifilm Corp | 画像分類装置および方法ならびにプログラム |
| JP2010273328A (ja) * | 2009-04-20 | 2010-12-02 | Fujifilm Corp | 画像処理装置、画像処理方法およびプログラム |
Also Published As
| Publication number | Publication date |
|---|---|
| US8737726B2 (en) | 2014-05-27 |
| CN102893304A (zh) | 2013-01-23 |
| JP5890325B2 (ja) | 2016-03-22 |
| JPWO2012102015A1 (ja) | 2014-06-30 |
| CN102893304B (zh) | 2016-02-10 |
| US20130004064A1 (en) | 2013-01-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102725756B (zh) | 图像分类装置、方法、程序、记录程序的记录介质及集成电路 | |
| JP5890325B2 (ja) | 画像データ処理装置、方法、プログラム及び集積回路 | |
| US9058342B2 (en) | Image classification device, method, program, recording media with program thereon, and integrated circuit | |
| US11676390B2 (en) | Machine-learning model, methods and systems for removal of unwanted people from photographs | |
| US7995106B2 (en) | Imaging apparatus with human extraction and voice analysis and control method thereof | |
| EP2567536B1 (en) | Generating a combined image from multiple images | |
| CN102576461B (zh) | 评估数字图像的美学质量 | |
| US8532347B2 (en) | Generation and usage of attractiveness scores | |
| JP5686673B2 (ja) | 画像処理装置、画像処理方法およびプログラム | |
| CN101262561B (zh) | 成像设备及其控制方法 | |
| JP4474885B2 (ja) | 画像分類装置及び画像分類プログラム | |
| JP4490214B2 (ja) | 電子アルバム表示システム、電子アルバム表示方法、及び電子アルバム表示プログラム | |
| JP5640621B2 (ja) | 赤目オブジェクト候補を分類する方法、コンピューター読み取り可能な媒体および画像処理装置 | |
| CN118212231A (zh) | 图像处理方法及相关设备 | |
| JP2006236218A (ja) | 電子アルバム表示システム、電子アルバム表示方法、及び電子アルバム表示プログラム | |
| JP2006079460A (ja) | 電子アルバム表示システム、電子アルバム表示方法、電子アルバム表示プログラム、画像分類装置、画像分類方法、及び画像分類プログラム | |
| JP2013149034A (ja) | 画像表示装置、画像表示方法及びプログラム | |
| JP2013197827A (ja) | 撮影補助装置、方法、並びに該撮影補助装置を搭載した撮影装置及びカメラ付き携帯端末 | |
| US20260011064A1 (en) | Image processing device, operation method of image processing device, and operation program of image processing device | |
| JP5267183B2 (ja) | 画像レイアウト装置およびプログラム | |
| JP2003208615A (ja) | 画像処理装置、及びプログラム |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| WWE | Wipo information: entry into national phase |
Ref document number: 201280001015.9 Country of ref document: CN |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 13578392 Country of ref document: US |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 12739331 Country of ref document: EP Kind code of ref document: A1 |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 2012554676 Country of ref document: JP |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 12739331 Country of ref document: EP Kind code of ref document: A1 |