JP2012212239A - Moving-body monitoring system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a moving-body monitoring system capable of effectively detecting the movement of an object with respect to an approach monitoring area and monitoring the moving-body with a small processing load.SOLUTION: A moving-body monitoring system comprises: an image extraction part for extracting a first image area which is the image area of a first indicator zone optionally set outside an approach monitoring area within a camera photographic field and a second image area which is the image area of a second indicator zone optionally set between the approach monitoring area and the first indicator zone from a photographed image; a feature amount calculation part for calculating first and second image feature data which are the respective image feature amounts of the first and second image areas; an evaluation data calculation part for calculating object movement evaluation data for evaluating object movement with respect to the approach monitoring area on the basis of the first and second image feature data; and an object movement estimation part for estimating one or both of the approach and exit of the object with respect to the approach monitoring area on the basis of the temporal change of the object movement evaluation data.

Description

  The present invention relates to a moving object monitoring system that monitors the movement of an object relative to an approach monitoring area using an image captured by a camera.

  Patent Document 1 discloses an image monitoring apparatus that monitors a moving object in a monitoring area by calculating a plurality of digitally captured images obtained using a monitoring camera. In this apparatus, a difference image of the input captured image is calculated, the calculated difference image is binarized to generate a difference binarized image, a change area is detected from the generated difference binarized image, The detected change area is used as a moving object, and the shading pattern of the change area is stored in a storage unit. The moving object can be tracked by extracting a region having a high degree of similarity and determining the moving object by comparing the gray pattern and the gray pattern in the storage means. That is, in order to analyze the movement of the moving object, the moving object is extracted from the captured image at a certain time, and the corresponding region of the moving object is extracted from the image at the subsequent time. Therefore, even if the position in the monitoring area of the moving object extracted first is a position that is not so important, there is a waste of tracking the moving object.

  Further, Patent Document 2 describes an entrance / exit detection device that detects the entrance / exit of a moving object by detecting the movement of a moving object including a human using the difference calculation result of the captured image density. This device is an entrance / exit detection device that detects the entrance / exit of a moving object by detecting the movement of a moving object including a person, and includes processing image data input from the image input device and input of the processing image data. Based on the analysis data, a difference detection unit that obtains difference data from the reference image data that has been input and held, an image processing unit that executes predetermined data analysis processing based on the difference data, and extracts analysis data A position detection unit for detecting the position of the moving object, a motion detection unit for detecting the movement of the moving object based on the analysis data or the position information of the processed image data processed in tandem with each other, Based on the position information detected by the position detector and the motion information detected by the motion detector, the movement in the entry direction, the movement in the exit direction within the imaging range of the image input device, and between the entrance and exit directions And a entry and exit determination portion determines the entry and exit of the moving object to distinguish U-turn movement. Here, the difference detection unit holds the image data input from the image input device, the processed image data input latest among the input image data, and input and held before that. Difference data with reference image data is generated. In the “initial image input mode”, the reference image data is captured by the image input device, input to the difference detection unit, and the initial image stored in the image memory in the difference detection unit is used. In the “initial image input mode”, the operation after obtaining the reference image data is not performed. After changing the mode from the “initial image input mode” to the “operation mode”, the difference detection unit calculates an absolute difference between the reference image data acquired in the “initial image input mode” and the latest input processed image data. Difference data obtained by calculating the value is generated and output to the image processing unit. Therefore, in this apparatus as well, since difference data between the captured image input every time and the reference image input and held in advance is obtained, the processing load for detecting the motion of the moving object increases.

Japanese Patent Laid-Open No. 6-231252 (paragraph number [0016-0024], FIG. 2) Japanese Patent Laying-Open No. 2006-107060 (paragraph number [0011-0053], FIG. 1)

  In view of the above circumstances, there is a demand for a moving object monitoring system that can effectively detect the movement of an object relative to a predetermined approach monitoring area and monitor the moving object with a small processing burden.

  The moving object monitoring system according to the present invention for monitoring the movement of an object with respect to an approach monitoring area using a photographed image by a camera is a first index band arbitrarily set outside the approach monitoring area within the photographing field of view of the camera. A first image area that is an image area of the second index band is extracted from the captured image and is arbitrarily set between the approach monitoring area and the first index band within the field of view of the camera. An image extraction unit that extracts a second image area that is an area from the captured image, calculates first image feature data that is an image feature quantity of the first image area, and uses the image feature quantity of the second image area. A feature amount calculation unit for calculating certain second image feature data, and for evaluating object movement with respect to the approach monitoring area based on the first image feature data and the second image feature data. An evaluation data calculation unit that calculates body movement evaluation data, and an object movement estimation unit that estimates the entry and / or exit of an object to and from the entry monitoring area based on a change over time of the object movement evaluation data. Yes.

  According to this configuration, with respect to the evaluation of the object movement, the entire captured image is not processed, but only the image area of the index band set in advance at an arbitrary position around the approach monitoring area is processed. The image feature amount is calculated from the image area extracted from the captured image, that is, the first image area and the second image area, and the image feature amount is calculated as an input parameter, thereby monitoring the approach. The object movement evaluation data for evaluating the object movement with respect to the area is derived. The object movement evaluation data is sequentially derived with time, and the approach and / or exit of the object to / from the monitoring area is estimated from the change in the object movement evaluation data with time. As the image feature amount handled here, an image feature amount that changes due to the presence of an object between the camera and the index band is suitable. Examples of such image feature amounts include pixel density values, node histograms, color distribution values, and the like. Object movement evaluation data as information on the presence or absence of an object in the vicinity of the approach monitoring area at the time of photographing by comparing the image feature amounts of the image areas corresponding to the first index band and the second index band, respectively. Is obtained. Therefore, the object movement evaluation data is obtained over time, and from the change over time, the object enters the approach monitoring area via the first and second indicator bands, or the first and second from the approach monitoring area. It can be estimated that the vehicle exits through the indicator band. In addition, since the position of the first image area and the second image area is specified by the index body provided for the approach monitoring area, the first image area and the second image area to be monitored by the installation of the index body are determined. Can be identified.

  In the present invention, the first image area and the second image area of the captured image corresponding to the first index band and the second index band set around the approach monitoring area are the processing targets. The positions of the first image area and the second image area are important. For this reason, in one preferred embodiment of the present invention, the first index band member that defines the first index band and the second index band member that defines the second index band are the floor surface of the approach monitoring area. The position of the first image area and the second image area in the captured image is set from the first index band member and the second index band member reflected in the captured image. An index position setting unit is provided. In this configuration, before executing the moving object monitoring, the second indicator band member is disposed inside the periphery of the approach monitoring area where it is necessary to monitor the intrusion of the object, and the first indicator band member is disposed outside the second index band member. In this state, from the captured image including the first index band member and the second index band member, the position of the first index band member (position of the first image region) and the position of the second index band member in the captured image (The position of the second image area) is set in advance. By setting the position in advance, the first image area and the second image area can be extracted from the captured image using the position. Moreover, if it comprises a member like a tape affixed to a floor surface (or road surface), an indicator belt member will become simple and the change of an entrance monitoring area can be responded easily.

  In order to prevent the first image region and the second image region in the captured image from becoming unclear due to illumination or other imaging conditions, the first index band member and the second index band member for the camera being used. Must be easy to project. For this reason, in a preferred embodiment of the present invention, the first index band member and the second index band member are made of a retroreflective material or a fluorescent material. When there is an illumination light source on the camera side, the retroreflective material has a function to return the incident light straight in the direction of the light source, so that it is possible to return the illumination light irradiated from the camera side to the camera satisfactorily. A first image area and a second image area are obtained. In addition, when there is no illumination light source on the camera side, a clear first image region and second image region are obtained, and when blocked by a moving object, the influence of the object can be detected with a large brightness difference.

  Since a captured image acquired by a general camera represents a density value, which is a degree of brightness, in units of pixels, an image feature amount calculated from the first image region or the second image region is set for each image feature amount. It is convenient to use the density distribution of the image area. For example, since the density distribution changes depending on whether or not an object exists in the space between the first index band or the second index band and the camera, the image feature data indicating this density distribution is substantially As it is, it can be used as object movement evaluation data for evaluating object movement with respect to the approach monitoring area.

  Assuming that the first image area and the second image area extend in a predetermined direction in parallel to each other, the image feature data has a more specific form, preferably for each section in which the image area is divided in the predetermined direction. It can be generated as comprising density values. Here, the phrase “extending in parallel in a predetermined direction” means that the distance between them is substantially equal, and the first image area and the second image area are coaxially arranged on the outer peripheral side segment. And an inner peripheral segment.

  In one preferred embodiment of the present invention, the object movement evaluation data is difference value data between the first image feature data and the second image feature data. If the object exists between the first index band and the second index band, which index band the object has a large influence from the difference value data between the first image feature data and the second image feature data, That is, information indicating which index band is approaching can be obtained from the difference value data.

It is the model which illustrated the detection principle of the moving object in the periphery of an approach monitoring area. It is a schematic diagram which shows an example of the whole structure of the moving object monitoring system by this invention, and its functional block. It is explanatory drawing illustrating the moving object detection in a moving object monitoring system when a person enters into an approach monitoring area. It is explanatory drawing which illustrates the time-dependent fluctuation | variation of the moving object detection in a moving object monitoring system when a person enters the approach monitoring area.

Before describing an embodiment of a moving object monitoring system according to the present invention, the basic principle of moving object monitoring employed in this moving object monitoring system will be described with reference to FIG.
First, in this moving object monitoring system, an index band is set around the outside of a specific area (herein referred to as an entry monitoring area) where it is necessary to monitor the approach of a moving object (# 01). At least two index bands are set in the approach direction to the approach monitoring area. In the example of FIG. 1, two of the first index band far from the approach monitoring area and the second index band nearer are set. That is, the second index band is set between the first index band and the approach monitoring area. When an object enters the approach monitoring area from the outside, the second index band passes through the first index band, and then the second index band. It will pass through the indicator band and reach the approach monitoring area. The camera that acquires the captured image processed by the moving object monitoring system is arranged at an upper position so that the first index band, the second index band, and the approach monitoring area are within the imaging field of view (# 02). . The approach monitoring area may be partially within the field of view of the image, not all of it. Basically, since the index band is arranged so that the moving object that enters is positioned between the camera and the index band, it is convenient to set it on the boundary surface of the monitoring space such as the floor surface. The index band may be a member such as a tape that can be freely attached to the floor surface, or may be a virtual section logically defined on the floor surface that does not actually exist. .

  The captured image acquired by the cameras arranged in this way includes the first index band and the second index band. In order to easily extract the first image area and the second image area, which are the image areas of the first index band and the second index band in the captured image, from the captured image, the coordinate positions in the captured image are calculated. Therefore, as preprocessing, the operator selects the first image region and the second image region while viewing the captured image, and the coordinate position is registered as image extraction position information. The coordinate positions registered here are valid as long as the position / posture of the camera and the positions of both index bands are not changed.

  When monitoring processing by the moving object monitoring system is executed, captured images are sent from the camera over time. A first image area that is an image area of the first index band and a second image area that is an image area of the second index band are extracted from the input captured image (# 03). First image feature data that is an image feature amount of the first image region is calculated, and second image feature data that is an image feature amount of the second image region is calculated (# 04). The image feature amount handled here may be any as long as it changes when the index band is blocked by the moving object with respect to the shooting direction of the camera. A typical example is a collection of density values, also called luminance values of each pixel, that is, a density distribution. In this case, the density distribution may be a collection of density values in position units of the image area, or may be a density histogram. Also, depending on the moving object, color information such as hue can be used as the image feature amount. Furthermore, if the image feature of the object to be monitored is determined, an image feature amount suitable for it may be adopted. This is because when monitoring an object such as a human, the distribution of density pixel values corresponding to the spatial position is suitable, but a density histogram may be suitable for gases such as city gas and carbon dioxide.

  The first image feature data obtained here is an image feature quantity of the first image area, that is, the first index band, and the second image feature data is an image feature quantity of the second image area, that is, the second index band, In addition, since the first index band and the second index band are arranged in a perspective relationship with respect to the approach monitoring area, an object of the approach monitoring area is based on the first image feature data and the second image feature data. Travel can be evaluated. For example, when the first index band is blocked by an object and the second index band is blocked by the object, the first index feature data is compared with the second image feature data to affect the first index band. The influence of moving objects can be calculated. That is, it is estimated that the moving object has entered the first index zone. After that, when the first index band and the second index band are obstructed by the object, the influence degree of the moving object exerted on the first index band and the second index band becomes almost the same, so the first index band enters the first index band. It is estimated that the moving object has further entered the second index zone. When the first image feature data and the second image feature data are calculated in this way, the object movement evaluation data for evaluating the object movement with respect to the approach monitoring area can be obtained from the calculation result (# 05).

One simple method for obtaining object movement evaluation data from the first image feature data and the second image feature data is to take a difference between the first image feature data and the second image feature data, and to calculate the difference value data. The object movement evaluation data is used, and the entry of the moving object into the entry monitoring area can be evaluated by evaluating the object movement evaluation data with time. Furthermore, a filtering process may be introduced in order to reduce erroneous evaluation due to noise included in data collection (environmental light change, movement of minute objects, etc.).
For example, threshold determination or size filtering is applied to the first difference value data that is the difference value between the first image feature data obtained at time: t and the first image feature data obtained at time: t + Δt. 1 Find feature data. Similarly, threshold determination or size filtering is applied to the second difference value data, which is a difference value between the second image feature data obtained at time: t and the second image feature data obtained at time: t + Δt, and time-dependent. Second feature data is obtained. The object movement evaluation data may be obtained through the calculation of the difference or the like from the first temporal feature data and the second temporal feature data obtained over time in this way. The object movement evaluation data sequentially obtained in this way can be further evaluated over time to determine the approach direction of the moving object.
In addition, as a noise reduction measure when calculating image features, statistical calculations such as smoothing filter and sampling time adjustment, intermediate value calculation and moving average for pixel values of multiple captured images acquired with short sampling time It is also possible to calculate a representative value and calculate an image feature amount based on a photographed image made up of the representative value.
The present invention includes performing various filtering processes known in the image processing field other than the above for noise reduction on the acquired captured image, image feature data, and evaluation data.
Further, when the size of the moving object to be evaluated is known in advance, an exclusion process can be performed in which an allowable range based on the size is set and data outside the allowable range is removed.

  Since the object movement evaluation data obtained in this way includes the position information of the moving object at that time, the object enters or exits the entry monitoring area based on the temporal change of the object movement evaluation data. Alternatively, both can be estimated. For example, as schematically shown in FIG. 1, it is estimated from the object movement evaluation data at the time point t0 that there is no object blocking the first index band and the second index band, and the object movement evaluation at the time point t1. It is estimated from the data that the object is blocking the first index band, and it is estimated from the object movement evaluation data at time t2 that the object is blocking the second index band, and from the object movement evaluation data at time t3 If it is estimated that there is no object blocking the first index band and the second index band, it can be estimated that the moving object has entered the approach monitoring area. From this, a warning can be notified at time t3.

  One embodiment of the moving object monitoring system according to the present invention, which operates based on the above-mentioned basic principle of moving object monitoring, will be described below. FIG. 2 is a schematic diagram showing the overall configuration and functional blocks of this moving object monitoring system. In this system, the area for placing valuables displayed in the exhibition room is set as the entrance monitoring area X, and the ceiling of the exhibition room is shown. The digital camera (hereinafter simply referred to as a camera) 10 installed in the camera 10 has its shooting direction directed to the floor surface, and a placement area and its peripheral area are placed in the shooting field of view. The moving object is a person. Further, in this embodiment, an index tape member Z in the form of an adhesive tape that is detachably attached to the floor surface is used as the index band. The index band member Z is composed of a first index band member Z1 and a second index band member Z2, so that the second index band member Z2 is positioned between the first index band member Z1 and the approach monitoring area X. It is arranged around the approach monitoring area X. Therefore, in order for a person to enter the approach monitoring area X, the first index band member Z1 is first exceeded, and then the second index band member Z2 is exceeded. In this example, since entry is possible from the entire periphery of the entry monitoring area X, an index band member set including the first index band member Z1 and the second index band member Z2 is arranged in four directions. However, for the sake of simplicity, it is assumed that there is only one index band member set thereafter.

  A core component of the moving object monitoring system is a monitoring controller 20 that receives a captured image acquired by the camera 10, and substantially includes a computer. In addition to the camera 10, an input operation device 11 such as a mouse and a keyboard, a monitor 12, and a device controller 40 are connected to the monitoring controller 20. The device controller 40 controls the operation of the operation device 41 such as an alarm device or a security device based on the moving object detection information output when the movement of the object to the approach monitoring area X is detected.

  The monitoring controller 20 includes an image input unit 21, an index position setting unit 22, a GUI (graphic user interface) unit 23, an object movement estimation unit 24, a device control unit 25, and an image processing unit 50. The image input unit 21 develops captured images sent from the camera 10 at predetermined time intervals in a memory. The GUI unit 23 displays the operation on the monitor 12 as a graphic in order to simplify the operation of the operator using the input operation device. As the preprocessing, the index position setting unit 22 performs an operation in which the operator designates the contour while confirming the images of the first index band member Z1 and the second index band member Z2 reflected in the captured image on the monitor screen. The coordinate positions in the captured image of the image area (first image area) in the captured image of the first index band member Z1 and the image area (second image area) in the captured image of the second index band member Z2 are set. The position information that is the coordinate position set here is given to the image processing unit 50. When the image area of the index band member Z is easily distinguishable from other areas and can be easily detected by an edge detection filter or the like, the first image area and the second image area in the captured image are automatically included in the captured image. It is also possible to adopt a method of setting the coordinate position.

  The image processing unit 50 includes a preprocessing unit 51, an image extraction unit 52, a feature amount calculation unit 53, and an evaluation data calculation unit 54. The pre-processing unit 51 performs pre-processing such as white balance, density adjustment, resolution conversion, and rotation / reduction processing as necessary on the captured image developed in the memory. Based on the position information received from the index position setting unit 22, the image extraction unit 52 extracts a first image area that is an image area of the first index band Z1 from the captured image, and an image area of the second index band Z2. The second image area is extracted.

  The feature amount calculation unit 53 calculates first image feature data that is an image feature amount of the first image region, and calculates second image feature data that is an image feature amount of the second image region. Here, the density gradation value which is the pixel value of the pixels constituting the image area is used as it is as the image feature amount. However, in order to simplify the subsequent processing, each image region is subjected to conversion for reducing the number of pixels and gradation reduction conversion for simplifying the gradation value by averaging processing (mosaic processing). This conversion process reduces the number of pixels in each image area and also reduces the pixel value (density gradation value). For example, when the image feature data is changed from 8 bits to 3 bits or 2 bits (binarized data), the image feature data can have a small calculation load.

  The evaluation data calculation unit 54 calculates object movement evaluation data for evaluating object movement with respect to the approach monitoring area X based on the first image feature data and the second image feature data. Here, the difference between the first image feature data and the second image feature data is adopted as the simplest method. By this difference, a person can detect the passing state of the first index band member Z1 or the second index band member Z2 in a state where disturbance is suppressed. The object movement evaluation data obtained by the difference between the first image feature data and the second image feature data is used as it is to evaluate the movement of the object passing through the first index band Z1 and passing through the second index band Z2. However, for simpler evaluation, the average value or integrated value of the pixel values of the obtained image feature data may be used as the object movement evaluation data.

  The object movement estimation unit 24 estimates a person's entry and / or exit from the entry monitoring area X based on the temporal change of the object movement evaluation data. That is, the captured images sent from the camera 10 at predetermined time intervals are sequentially subjected to image processing by the preprocessing unit 51, the image extraction unit 52, and the feature amount calculation unit 53, so that the object movement evaluation data at the predetermined time intervals is obtained. Is output from the evaluation data calculation unit 54. By evaluating a value group constituting the output object movement evaluation data or a representative value (average value or integrated value) of the value group, it is possible to determine whether the person has entered the entry monitoring area X.

  The device control unit 25 receives the entry information output when the object movement estimation unit 24 determines that the person has entered the entry monitoring area X, and sends the control information to the device controller 40 so as to perform necessary treatment. The device controller 40 selects a necessary countermeasure from pre-registered countermeasure scenarios based on the sent control information, and controls each operating device 41 according to the countermeasure scenario.

  Next, the operation of the image processing unit 30 and the object movement estimation unit 24 when a person enters the entry monitoring area X will be described with reference to FIGS. 3 and 4 which are simplified illustrations for the purpose of explanation. explain.

(1) A person exists further outside the first indicator band member Z1 [see FIG. 3 (a)]
Since the first image region and the second image region extracted from the photographed image acquired at this time t0 are not affected by a person, all the pixel values of the first image feature data and the second image feature data are the same. It is “7”. Here, the number of pixels in each image area is 5, and the pixel value is a 3-bit gradation density. The five pixel values of the object movement evaluation data obtained by subtracting the second image feature data from the first image feature data are all “0”. When the object movement evaluation data is an integrated value of the subtraction value, the value is “0”. From this, it is estimated that no person (object) exists on the first index band member Z1 and the second index band member Z1.

(2) A person is passing over the first indicator band member Z1 [see (b) of FIG. 3]
The first image area extracted from the photographed image acquired at this time t1 is affected by a person, and in this example, it is dark in the central area. Accordingly, the pixel value of the first image feature data is decreased, and the pixel values are “7”, “4”, “0”, “4”, “7” in order from the end. This is because the amount of light incident on the camera 10 is set so that the first indicator band member Z1 is larger than the person (or clothes). For this reason, it is convenient to use a retroreflective material or a light-emitting material as the material of the first index band member Z1, or to apply a bright paint. On the other hand, the second image area is not yet influenced by humans, and the pixel values of all the pixel values of the second image feature data are “7”. Accordingly, the pixel values of the object movement evaluation data are “0”, “−3”, “−7”, “−3”, and “0”, and the integrated value is “−13”. From this, it is presumed that some kind of person exists on the first index band member Z1.

(3) A person is passing over the first index band member Z1 and the second index band member Z2 [see FIG. 3 (c)]
The first image region and the second image region extracted from the photographed image acquired at this time t2 are affected by a person, and in this example, both are dark in the central region. Accordingly, the pixel values of the first image feature data and the first image feature data are lowered, and the pixel values of the first image feature data are “7”, “5”, “1”, “5” in order from the end. “7”, and in the second image feature data, “7”, “7”, “3”, “6”, “7”. Accordingly, the pixel values of the object movement evaluation data are “0”, “−2”, “−2”, “−1”, “0”, and the integrated value is “−5”. From this, it is estimated that there is some person on the first index band member Z1 and the second index band member Z1, more specifically, the first index band member Z1 is slightly inclined. Is done.

(4) A person is passing over the second index strip member Z1 (see (d) of FIG. 3)
The second image area extracted from the photographed image acquired at this time t3 is affected by a person, and here, since it is dark in the central area, the pixel value of the second image feature data is reduced. The pixel values are “7”, “4”, “0”, “4”, “7” in order from the end. On the other hand, the first image area is no longer affected by humans, and the pixel values of all the pixel values of the first image feature data are “7”. Therefore, the pixel values of the object movement evaluation data are “0”, “3”, “7”, “3”, “0”, and the integrated value is “13”. From this, it is presumed that some kind of person exists on the second index band member Z1.

(5) A person enters the entry monitoring area X beyond the second index belt member Z1 (see (e) of FIG. 3).
Since both the first image region and the second image region extracted from the photographed image acquired at this time t4 are no longer influenced by humans, the first image feature data and the second image feature data are all The pixel value is “7”, and the integrated value is “0”. From this, it is estimated that there is no person on the first index band member Z1 and the second index band member Z1.

  As shown in FIG. 4, when the human presence estimation results obtained by the processes (1) to (5) above ((a) to (e) in FIG. 3) are viewed over time, (1) non- Existence (2) Existence on the first indicator band member (3) Existence on the first and second indicator band members (4) Existence on the second indicator band member (5) Absence of evaluation results over time ((A) to (e) in FIG. 4). From this, it can be estimated that a person has entered the approach monitoring area X from the outside. Here, if the processing result of (3) shifts to the processing result of (1) through (2), it can be estimated that the person who has entered once made a U-turn. Furthermore, conversely, when the processing results from (5) to (1) occur, it can be estimated that a person has left the entry monitoring area X.

  In the description of the data handled by the image processing unit 30 and the object movement estimation unit 24 using FIG. 3, the data is 1 × 6 pixels for the sake of simplicity, but actually, data of a larger size is handled. Is preferred. Furthermore, it is preferable that the captured image capturing period from the camera 10 is 1 second or less, and object movement evaluation data of several tens of times before an object including a person passes through the second index band member from the first index band member. Can be calculated, whereby the movement of the object from the first index band member to the second index band member can be estimated with high resolution.

[Another embodiment]
(1) In the description of the above embodiment, the evaluation data calculation unit 54 calculates the object movement evaluation data by taking the difference between the first feature value data and the second feature value data at the same time. It is also possible to apply a temporal method to the algorithm.
For example, the noise required for the first difference value data that is the difference value between the first image feature data obtained previously (time: t) and the first image feature data obtained thereafter (time: t + Δt). A removal process is performed to obtain first characteristic data over time. Similarly, it is necessary for the second difference value data which is a difference value between the second image feature data obtained previously (time: t) and the second image feature data obtained thereafter (time: t + Δt). A noise removal process is performed to obtain second characteristic data over time. The object movement evaluation data may be obtained through the calculation of the difference or the like from the first temporal feature data and the second temporal feature data obtained over time in this way. The object movement evaluation data sequentially obtained in this way is sent to the object movement estimation unit 24, where it is further evaluated over time to determine the approaching direction of the moving object.
As yet another form, the evaluation data calculation unit 54 and the object movement estimation unit 24 are integrated, and the difference between the first image feature data obtained at time: t and the first image feature data obtained at time: t + Δt. First evaluation value data is obtained by performing threshold judgment or size filtering on the first difference value data that is a value. Similarly, threshold determination or size filtering is applied to the second difference value data that is the difference value between the second image feature data obtained at time: t and the second image feature data obtained at time: t + Δt, and the second difference is obtained. Obtain evaluation data. In this way, the first evaluation data group and the second image feature data group obtained over time can be evaluated to determine the approach direction of the moving object.
As described above, the evaluation data calculation unit 54 and the object movement estimation unit 24 of the present invention have a function of realizing evaluation data calculation and object movement estimation by a free method. What is important in the present invention is that the evaluation data calculation unit 54 and the object movement estimation unit 24 have various modes for evaluating the object movement with respect to the approach monitoring area based on the first image feature data and the second image feature data. The object movement evaluation data is calculated, and the approach and / or exit of the object to / from the approach monitoring area is estimated based on the temporal change of the object movement evaluation data.
(2) This moving object monitoring system is not only for safety approach monitoring, but also a guidance system for notifying guidance information when a person approaches, or an automatic lighting system for lighting when a person enters a room Can also be used. In addition, by using a radiating member that reflects invisible light on the index band member Z and a non-visible light camera that can capture the index band member Z, it is possible to perform monitoring without being conscious of humans at all. is there.

  The present invention can be applied to a system that estimates movements of various objects through changes in image feature values.

10: Camera 20: Monitoring controller 22: Index position setting unit 24: Object movement estimation unit 25: Device control unit 30: Image processing unit 32: Image extraction unit 33: Feature amount calculation unit 34: Evaluation data calculation unit 40: Device controller Z: Index band (index band member)
Z1: First index band (first index band member)
Z2: Second index band (second index band member)
X: Approach monitoring area

Claims (6)

A moving object monitoring system that monitors the movement of an object relative to an approach monitoring area using an image captured by a camera,
A first image area, which is an image area of a first index band arbitrarily set outside the entry monitoring area within the photographing field of view of the camera, is extracted from the photographed image, and the approach is entered within the photographing field of the camera. An image extraction unit that extracts from the captured image a second image area that is an image area of a second index band arbitrarily set between a monitoring area and the first index band;
A feature amount calculator that calculates first image feature data that is an image feature amount of the first image region, and that calculates second image feature data that is an image feature amount of the second image region;
An evaluation data calculation unit for calculating object movement evaluation data for evaluating object movement with respect to the approach monitoring area based on the first image feature data and the second image feature data;
An object movement estimator that estimates the entry and / or exit of an object to and from the entry monitoring area based on a change over time of the object movement evaluation data;
A moving object monitoring system.   The first index band member that defines the first index band and the second index band member that defines the second index band are members that can be arbitrarily installed on the floor surface of the approach monitoring area, and are reflected in the captured image. The index position setting part which sets the position of the 1st image field and the 2nd image field in the picked-up image from the 1st index belt member and the 2nd index belt member which were included is provided. Moving object monitoring system.   The moving object monitoring system according to claim 2, wherein the first index band member and the second index band member are made of a retroreflective material or a fluorescent material.   The moving object monitoring system according to claim 1, wherein the image feature amount is a density distribution of the image region.   The first image area and the second image area extend in a predetermined direction in parallel with each other, and the first image feature data and the second image feature data are provided for each section obtained by dividing the image area in the predetermined direction. The moving object monitoring system according to claim 4, wherein the moving object monitoring system is a density value.   The moving object monitoring system according to claim 5, wherein the object movement evaluation data is difference value data between the first image feature data and the second image feature data.

Images