JP2013149039A - Vehicle color determination device, computer program, and vehicle color determination method - Google Patents

Abstract

A vehicle color determination device, a computer program, and a vehicle color determination method capable of accurately determining the color of a vehicle.
A license plate detector detects a license plate area from a captured image. The license plate luminance value calculation unit 13 calculates the luminance value of a specific part of the license plate region. The vehicle body color determination area specifying unit 15 specifies a determination area for determining the color of the vehicle based on the position of the license plate area on the captured image. The color sample table selection unit 16 selects a color sample table corresponding to the luminance value of a specific part of the license plate area. The vehicle body color determination unit 18 determines the color of the vehicle based on the color component of each pixel in the determination region and the color component of the selected color sample table.
[Selection] Figure 1

Description

  The present invention relates to a vehicle color determination device that determines the color of a vehicle based on a captured image obtained by imaging the vehicle, a computer program for realizing the vehicle color determination device, and a vehicle color determination method.

  A technology has been developed to determine the travel time required to travel between predetermined points by identifying the number plates (vehicle numbers) of vehicles traveling on the road at multiple points and collating the identified number plates. ing.

  On the other hand, there has been a problem that some characters cannot be recognized due to bending or dirt of the license plate. Therefore, color image data from a camera installed on the road is acquired, a discrimination area for discriminating the color of the vehicle such as a vehicle bonnet is extracted from the obtained color image data, and the color of the extracted discrimination area is identified Discloses a vehicle color discrimination device that discriminates the color of a vehicle (see Patent Document 1).

JP 2000-222673 A

  For example, the technique disclosed in Patent Document 1 determines the color of a vehicle based on R (red), G (green), and B (blue) values (or values converted into another color system) of a color image. Yes. However, even for vehicles of the same color, the RGB values of the captured images differ depending on the sunlight conditions (illumination conditions) such as the weather at the time of imaging of the vehicle, the time zone, the sun or shade, and the surrounding environment. May be wrongly determined.

  The present invention has been made in view of such circumstances, and provides a vehicle color determination device capable of accurately determining the color of a vehicle, a computer program and a vehicle color determination method for realizing the vehicle color determination device. The purpose is to provide.

  A vehicle color determination device according to a first aspect of the present invention is a vehicle color determination device that determines the color of a vehicle based on a captured image obtained by imaging the vehicle, based on the pixel value of each pixel constituting the captured image. Detecting means for detecting a license plate area; luminance value calculating means for calculating a luminance value of a specific part of the license plate area detected by the detecting means; and a plurality of color components of the pixel corresponding to each of the plurality of luminance values; Storage means for storing a plurality of color sample tables associated with the colors of the pixels, selection means for selecting a color sample table corresponding to the luminance value calculated by the luminance value calculating means, and a license plate area detected by the detecting means Specifying means for specifying a determination region for determining the color of the vehicle based on the position on the captured image, the color component of each pixel of the determination region specified by the specifying means, and the selection means Characterized in that it comprises a determination means for determining the color of the vehicle based on the the color sample table of color components.

  According to a second aspect of the present invention, there is provided a vehicle color determination apparatus according to the first aspect, further comprising a counting unit that counts, for each color, the number of pixels in which the color component of each pixel in the determination area matches the color component of the color sample table. The determination means is configured to determine the color of the pixel as the color of the vehicle when the number of pixels counted by the counting means is equal to or greater than a threshold value.

  According to a third aspect of the present invention, in the vehicle color determination apparatus according to the first aspect, the storage means stores L * a * b * values corresponding to the colors of the color sample table. Calculation means for calculating the L * a * b * value of each pixel based on the color component, L * a * b * value calculated by the calculation means, and L * a * b * for each color in the color sample table An approximation degree calculating means for calculating the degree of approximation with respect to each pixel in the determination region; and a counting means for counting the number of pixels for which the degree of approximation calculated by the approximation degree calculating means is equal to or greater than a predetermined value, The determining means is configured to determine the color of the pixel as the color of the vehicle when the number of pixels counted by the counting means is equal to or greater than a threshold value.

  According to a fourth aspect of the present invention, in the vehicle color determination device according to any one of the first to third aspects, the bright portion area in which the luminance value of each pixel of the license plate area detected by the detection unit is greater than a luminance threshold value. An extraction means for extracting as a specific part is provided.

  According to a fifth aspect of the present invention, there is provided the vehicle color determination apparatus according to the fourth aspect, wherein when the luminance value of the bright portion area is equal to or higher than a predetermined upper limit value, the extraction means has a luminance value of each pixel of the license plate area. A dark part area smaller than a luminance threshold is extracted as the specific part.

  A vehicle color determination device according to a sixth aspect of the present invention is the vehicle color determination device according to any one of the first to fifth aspects, further comprising pixel extraction means for extracting an edge pixel based on a pixel value of each pixel in the determination region, The means is configured to exclude edge pixels extracted by the pixel extraction means from the specified determination area.

  According to a seventh aspect of the present invention, in any one of the first to sixth aspects, the vehicle color determination device extracts a gradation region in which the change in luminance value is uniform based on the luminance value of each pixel in the determination region. An area extracting unit is provided, and the specifying unit is configured to exclude the gradation area extracted by the area extracting unit from the specified determination area.

  A vehicle color determination device according to an eighth aspect of the present invention is the vehicle color determination device for determining the color of a vehicle based on a captured image obtained by imaging the vehicle, based on the pixel value of each pixel constituting the captured image. Detecting means for detecting a license plate area; luminance value calculating means for calculating a luminance value of a specific part of the license plate area detected by the detecting means; a plurality of color components of the pixel corresponding to a predetermined luminance value; Storage means for storing a color sample table in which the colors of the pixels are associated; and specifying means for specifying a determination area for determining the color of the vehicle based on the position of the license plate area detected by the detection means on the captured image; Correction means for correcting the luminance value of each pixel in the determination area according to the difference between the luminance value calculated by the luminance value calculation means and the predetermined luminance value; and each pixel in the determination area corrected by the correction means And color components, characterized in that it comprises a determination means for determining the color of the vehicle based on the color component of the color sample table.

  A computer program according to a ninth aspect of the invention is a computer program for causing a computer to determine the color of a vehicle based on a captured image obtained by imaging the vehicle. Detecting a license plate area based on the value; calculating a luminance value of a specific part of the detected license plate area; and a plurality of color components of the pixel and a color of the pixel corresponding to each of the plurality of luminance values Selecting a color sample table corresponding to the calculated luminance value from a plurality of color sample tables associated with each other, and determining the color of the vehicle based on the position of the detected license plate area on the captured image A step of specifying the determination region, a color component of each pixel of the specified determination region, and a color component of the selected color sample table Characterized in that to execute determining a color of the vehicle based on.

  A computer program according to a tenth aspect of the present invention is a computer program for causing a computer to determine the color of a vehicle based on a captured image obtained by capturing the vehicle, and for each pixel constituting the captured image to the computer. Detecting a license plate area based on the value; calculating a luminance value of a specific part of the detected license plate area; and determining a color of the vehicle based on a position of the detected license plate area on the captured image Identifying a determination region for correcting, a step of correcting a luminance value of each pixel of the determination region according to a difference between the calculated luminance value and a predetermined luminance value, and a color component of each pixel of the corrected determination region And a plurality of color components of a pixel corresponding to a predetermined luminance value and a color component of a color sample table in which the color of the pixel is associated. Characterized in that to perform the determining the color of the vehicle Te.

  A vehicle color determination method according to an eleventh aspect of the invention is a vehicle color determination method by a vehicle color determination device that determines a color of a vehicle based on a captured image obtained by imaging the vehicle, and corresponds to each of a plurality of luminance values. Storing a plurality of color sample tables in which a plurality of color components of the pixel and the color of the pixel are associated, and detecting a license plate area based on a pixel value of each pixel constituting the captured image; A step of calculating a luminance value of a specific part of the license plate area, a step of selecting a color sample table corresponding to the calculated luminance value, and a position of the detected license plate area on the captured image. A vehicle color is determined based on a step of specifying a determination region for determining a color, a color component of each pixel in the specified determination region, and a color component of a selected color sample table Characterized in that it comprises a step.

  A vehicle color determination method according to a twelfth aspect of the present invention is a vehicle color determination method by a vehicle color determination device that determines the color of a vehicle based on a captured image obtained by imaging the vehicle, in accordance with a predetermined luminance value. Storing a color sample table in which a plurality of color components of a pixel and the color of the pixel are associated, detecting a license plate area based on a pixel value of each pixel constituting the captured image, and a detected number A step of calculating a luminance value of a specific part of the plate region, a step of specifying a determination region for determining the color of the vehicle based on the position of the detected license plate region on the captured image, and the calculated luminance value And correcting the luminance value of each pixel in the determination area according to the difference between the predetermined luminance value, the color component of each pixel in the corrected determination area, and the color component of the color sample table. There characterized in that it comprises a determining a color of the vehicle.

  In the first invention, the ninth invention, and the eleventh invention, the license plate region is detected based on the pixel value of each pixel constituting the captured image. For example, the license plate area detects the vehicle based on the difference between the image when the vehicle does not exist and the image when the vehicle exists, and cuts out a rectangular license plate portion based on the detected vehicle image. It can be detected by performing processing. The brightness value of the specific part of the detected license plate area is calculated. The specific part is a part used for selecting the color sample table according to the luminance. For example, the specific part may be a background part (also referred to as a bright part) excluding a character part such as a number (car number) or a character part (dark part). May also be referred to). The color of the background portion of the license plate is determined, for example, white (ordinary car etc.) or yellow (light car). The luminance of the specific part can be, for example, an average value of the luminance of the background portion (bright portion).

  The storage means stores a plurality of color sample tables in which a plurality of color components of pixels and the colors of the pixels are associated with each other in correspondence with a plurality of luminance values. The plurality of luminance values are luminance values corresponding to respective brightnesses such as when the captured image is dark as a whole or when it is bright as a whole due to differences in sunshine conditions (illumination conditions). A color sample table corresponding to each of the plurality of luminance values is stored. The plurality of color components are, for example, R (red), G (green), and B (blue) values (or values converted into other color systems, such as L * a * b * values). The pixel color is a determination color when determining the color of the vehicle, and is, for example, white, silver, gray, red, purple, yellow, blue, green, black, or the like. The determination color can also include the color of the background portion (white, yellow) or character portion (green, black) of the license plate. If the color components are RGB, and each of RGB can take 256 values, for example, the color of the pixel can be determined in correspondence with all combinations of values. The values of the color components in the color sample table associated with each of the plurality of luminance values (for example, 256 values for each of RGB) differ depending on the luminance value.

  A color sample table corresponding to the calculated luminance value is selected. For example, when the background portion of the license plate is white, the color sample table corresponding to the average value of the luminance of the white portion (the luminance value matches or is closest) is selected. The color of the license plate is predetermined as white, so by calculating the luminance value (average luminance value), how bright or dark the body color is due to sunshine conditions (lighting conditions), etc. Can be estimated, and a color sample corresponding to the estimated brightness (average value of luminance) can be selected.

  A determination area for determining the color of the vehicle is specified based on the position of the detected license plate area on the captured image. For example, when the vehicle is imaged from the front, a parallelogram-shaped region having a predetermined dimension along the road direction from the center of the upper edge of the detected license plate region (for example, a rectangular region) is set as the determination region. be able to. In the case of a vehicle head, the region is, for example, a region corresponding to a vehicle hood. When the vehicle is imaged from the rear, the detected license plate area (for example, a rectangular area) is enclosed, and the area excluding the license plate area in the rectangular area larger than the license plate area is set as the determination area. be able to. In the case of a vehicle rear, this area is, for example, an area around the license plate excluding the rear window.

  The color of the vehicle is determined based on the color component of each pixel in the specified determination area and the color component of the selected color sample table. For example, a color corresponding to or close to a color component (for example, RGB value) of an arbitrary pixel in the determination region is determined as the color of the pixel, and the color is determined for all the pixels in the determination region. . If the number of pixels of the same color is equal to or greater than a predetermined threshold, the color of the pixel is determined as the color of the vehicle. As described above, by selecting a color sample table according to the luminance value of a specific part of a license plate whose color is determined in advance, use a color sample table that takes into account the influence of sunlight conditions (lighting conditions) and the like in advance. Therefore, it is possible to accurately determine the color of the vehicle without being affected by the difference in sunlight conditions (lighting conditions) such as the weather at the time of imaging of the vehicle, the time zone, the sun or the surrounding environment such as shade. .

  In the second invention, the number of pixels in which the color component of each pixel in the determination region matches the color component of the color sample table is counted for each color, and when the number of counted pixels is equal to or greater than a threshold value, The color of the pixel is determined as the color of the vehicle. If the color components are RGB, and each of RGB can take 256 values, for example, the color of the pixel can be determined in correspondence with all combinations of values. If the number of pixels having the same color component is determined and the number of pixels with the same color is equal to or greater than the threshold, the color component of the color sample table is determined by determining the color of the pixel as the vehicle color. Since the color of the vehicle can be determined only by referring to it, the determination process can be simplified.

  In the third invention, the storage means stores in advance L * a * b * values corresponding to the respective colors (determination colors for determining the color of the vehicle) of the color sample table. Based on the color component (for example, RGB value) of each pixel in the determination area, the L * a * b * value of each pixel is calculated. The L * a * b * value can be obtained from the RGB value using a predetermined conversion formula, for example. The degree of approximation between the calculated L * a * b * value and the L * a * b * value of each color in the color sample table is calculated for each pixel in the determination region. The degree of approximation is an index indicating how close the L * a * b * value of the pixel in the determination area and the L * a * b * value of the color sample table are, and each L * a * b * When the sum of the squares of the differences is the smallest, it can be assumed that the degree of approximation is the largest. The number of pixels for which the calculated degree of approximation is greater than or equal to a predetermined value is counted, and if the counted number of pixels is greater than or equal to a threshold, the color of the pixel is determined as the color of the vehicle. Instead of a color sample table corresponding to all combinations of color components (for example, 256 is the cube value when RGB can take 256 values), L * a * corresponding to the determination color Since the color sample table having the b * value is stored, the storage capacity can be reduced.

  In the fourth invention, the bright area where the luminance value of each pixel of the detected license plate area is larger than the luminance threshold is extracted as the specific part. The color of the background part excluding the character part such as the number (car number) is defined, for example, white (ordinary car etc.) or yellow (light car), and the luminance value is larger (high) than the character part Therefore, it is possible to extract a bright area having a luminance higher than a predetermined luminance threshold as a specific part. Since the color of a specific part is determined in advance, the brightness of the specific part (for example, an average value) is obtained, so that the color of the vehicle can be adjusted to sunlight conditions (lighting) such as the surrounding environment such as weather, time of day, sun or shade. It is possible to estimate how much it has been affected by the difference in (condition) (whether it has become brighter or darker).

  In the fifth invention, when the brightness value of the extracted bright area is equal to or higher than the predetermined upper limit value, the dark area where the brightness value of each pixel in the license plate area is smaller than the brightness threshold is extracted as the specific part. For example, when the luminance value is represented by 0 to 255, the predetermined upper limit value can be a value of about 255 or 250 to 254. The dark area is, for example, a character portion of a license plate, and is green when the bright area is white, and black when the bright area is yellow. When the brightness value of the bright area is equal to or higher than the upper limit value, it can be determined that saturation (saturation) has occurred in the captured image. Therefore, the saturation is determined by setting the dark area as a specific part instead of the bright area. Even if a car has a problem, how much the color of the vehicle is affected by the difference in the sunshine conditions (lighting conditions) such as the weather, time of day, sun or shade, etc. ) Can be estimated.

  In the sixth invention, the edge pixel is extracted based on the pixel value of each pixel in the determination region, and the edge pixel extracted from the determination region is excluded. For extraction of edge pixels, for example, an existing image filter such as a sobel filter can be used. A vehicle bonnet or the like has almost no texture, whereas a light, emblem, or front grill has a texture. Further, a glass portion such as a window also has a texture due to light transmission or reflection. By performing edge extraction and removing (ignoring) pixels with a strong edge strength (parts with texture) from the determination region, it is possible to prevent erroneous determination of the color of the vehicle.

  In the seventh aspect of the invention, a gradation region with a uniform change in luminance value is extracted based on the luminance value of each pixel in the determination region. For example, if a luminance value is sampled at a predetermined interval along a predetermined direction in the determination region, for example, the vertical direction (Y-axis direction) of the captured image, and there is a luminance value that changes constantly among the sampled luminance values A region including a position where the luminance value is sampled can be extracted as a gradation region. On the surface facing up, such as a bonnet, the color of the sky may appear due to the reflection of light, appearing as a gradation, and the color of the vehicle may appear different. By removing (ignoring) the gradation area from the determination area, it is possible to prevent the vehicle color from being erroneously determined.

  In the eighth invention, the tenth invention and the twelfth invention, the license plate region is detected based on the pixel value of each pixel constituting the captured image. For example, the license plate area detects the vehicle based on the difference between the image when the vehicle does not exist and the image when the vehicle exists, and cuts out a rectangular license plate portion based on the detected vehicle image. It can be detected by performing processing. The brightness value of the specific part of the detected license plate area is calculated. The specific portion may be, for example, a background portion (also referred to as a bright portion) excluding a character portion such as a number (vehicle number) or a character portion (also referred to as a dark portion). The color of the background portion of the license plate is determined, for example, white (ordinary car etc.) or yellow (light car). The luminance of the specific part can be, for example, an average value of the luminance of the background portion (bright portion).

  The storage means stores a color sample table in which a plurality of color components of a pixel and the color of the pixel are associated with each other in correspondence with a predetermined luminance value. The predetermined luminance value is, for example, a luminance value corresponding to average or general brightness when the luminance of the entire captured image changes due to a difference in sunshine conditions (illumination conditions). The brightness value can be set. The plurality of color components are, for example, R (red), G (green), and B (blue) values (or values converted into other color systems, such as L * a * b * values). The pixel color is a determination color when determining the color of the vehicle, and is, for example, white, silver, gray, red, purple, yellow, blue, green, black, or the like. The determination color can also include the color of the background portion (white, yellow) or character portion (green, black) of the license plate. If the color components are RGB, and each of RGB can take 256 values, for example, the color of the pixel can be determined in correspondence with all combinations of values.

  A determination area for determining the color of the vehicle is specified based on the position of the detected license plate area on the captured image. For example, when the vehicle is imaged from the front, a parallelogram-shaped region having a predetermined dimension along the road direction from the center of the upper edge of the detected license plate region (for example, a rectangular region) is set as the determination region. be able to. In the case of a vehicle head, the region is, for example, a region corresponding to a vehicle hood. When the vehicle is imaged from the rear, the detected license plate area (for example, a rectangular area) is enclosed, and the area excluding the license plate area in the rectangular area larger than the license plate area is set as the determination area. be able to. In the case of a vehicle rear, this area is, for example, an area around the license plate excluding the rear window.

  The luminance value of each pixel in the determination area is corrected according to the difference between the calculated luminance value and a predetermined luminance value. For example, the luminance value of the entire determination area is corrected so that the calculated luminance value matches a predetermined luminance value. For example, when the calculated luminance value is smaller than a predetermined luminance value, that is, when the average luminance value of a specific part of the license plate area is smaller than the predetermined luminance value, the captured image is dark as a whole. And the luminance value is corrected so that the luminance of the determination area matches or approximates the luminance of the color sample table. Further, when the calculated luminance value is larger than the predetermined luminance value, that is, when the average luminance value of the specific part of the license plate area is larger than the predetermined luminance value, the captured image is bright as a whole, so the determination area And the luminance value is corrected so that the luminance of the determination area matches or approximates the luminance of the color sample table. As a result, the luminance of the determination region can be matched or brought close to the luminance of the color sample.

  The color of the vehicle is determined based on the color component of each pixel in the corrected determination area and the color component of the color sample table. For example, a color corresponding to or close to a color component (for example, RGB value) of an arbitrary pixel in the determination region is determined as the color of the pixel, and the color is determined for all the pixels in the determination region. . If the number of pixels of the same color is equal to or greater than a predetermined threshold, the color of the pixel is determined as the color of the vehicle. As described above, the luminance value of the determination region is corrected so that the luminance value of the specific part of the license plate whose color is determined in advance matches (or approaches) the predetermined luminance value. The brightness value of the judgment area can be matched or brought close to the brightness of the color sample by offsetting the influence of the lighting, and the sunlight conditions (lighting conditions) such as the weather at the time of imaging of the vehicle, the time zone, the sun or the surrounding environment such as the shade The color of the vehicle can be accurately determined without being affected by the difference.

  According to the present invention, the color of a vehicle can be determined with high accuracy.

1 is a block diagram illustrating an example of a configuration of a vehicle color determination device according to a first embodiment. It is a schematic diagram which shows an example of a license plate. It is a schematic diagram which shows an example of the vehicle body color determination area | region at the time of imaging a vehicle from the front. It is a schematic diagram which shows an example of the vehicle body color determination area | region at the time of imaging a vehicle from back. 3 is an explanatory diagram illustrating an example of a configuration of a color sample table according to Embodiment 1. FIG. It is explanatory drawing which shows the relationship between the color component (RGB value) of a color sample table, and a vehicle body color. It is explanatory drawing which shows an example of the brightness | luminance selection table for selecting a color sample table. It is explanatory drawing which shows an example of the extraction method of a gradation area | region. It is explanatory drawing which shows a mode that a gradation area | region is excluded from a determination area | region. 4 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device according to the first embodiment. FIG. 6 is a block diagram illustrating an example of a configuration of a vehicle color determination device according to a second embodiment. 10 is an explanatory diagram illustrating an example of a configuration of a color sample table according to Embodiment 2. FIG. It is explanatory drawing which shows the relationship between the color component (L * a * b * value) of a color sample table, and a vehicle body color. 6 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device according to the second embodiment. FIG. 10 is a block diagram illustrating an example of a configuration of a vehicle color determination device according to a third embodiment. 10 is a schematic diagram illustrating an example of luminance value correction according to Embodiment 3. FIG. 12 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device according to the third embodiment. FIG. 10 is a block diagram illustrating an example of a configuration of a vehicle color determination device according to a fourth embodiment. 14 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device according to the fourth embodiment. It is a flowchart which shows the process sequence of the light vehicle determination by the vehicle color determination apparatus of Embodiment 1-4.

(Embodiment 1)
Hereinafter, the present invention will be described with reference to the drawings illustrating embodiments. FIG. 1 is a block diagram illustrating an example of the configuration of the vehicle color determination apparatus 100 according to the first embodiment. An imaging device 200 is connected to the vehicle color determination device 100. The imaging device 200 may be configured to be built in the vehicle color determination device 100. The imaging device 200 images a vehicle traveling on a road and outputs color image data (captured image) obtained by imaging to the vehicle color determination device 100.

  The vehicle color determination apparatus 100 includes a control unit 10 that controls each unit in the apparatus, an interface unit 11, a license plate detection unit 12, a license plate luminance value calculation unit 13, a license plate color determination unit 14, and a vehicle body color determination region specifying unit 15. A color sample table selection unit 16, a pixel number counting unit 17, a vehicle body color determination unit 18, an edge pixel extraction unit 19, a gradation area extraction unit 20, a storage unit 21, and the like.

  The interface unit 11 acquires image data (captured image) obtained by imaging with the imaging apparatus 200, and stores the acquired image data in the storage unit 21 in association with the imaging time in units of frames.

  The license plate detection unit 12 extracts a captured image from the storage unit 21 and detects a vehicle (including a part of the vehicle) based on the extracted captured image. As the vehicle detection process, for example, the vehicle can be detected based on the difference between the captured image when the vehicle is not present and the captured image when the vehicle is present.

  The license plate detection unit 12 has a function as detection means, and detects a license plate region by performing a cutting process of a rectangular license plate part based on the detected vehicle image. The license plate area can be specified by coordinates on the captured image of the four corners of the rectangle.

  The license plate luminance value calculation unit 13 functions as a luminance value calculation unit, and calculates the luminance value of a specific part of the license plate region detected by the license plate detection unit 12. The specific part is a part used for selecting the color sample table according to the luminance.

  FIG. 2 is a schematic diagram showing an example of a license plate. The specific part may be a background part (also referred to as a bright part) excluding a character part such as a branch name of a license plate, a classification number, a car model symbol, a serial number, or a character part (also referred to as a dark part). For example, as shown in FIG. 2, when the vehicle type is an ordinary car, the background part (bright part) is white and the character part (dark part) is green. It is. When the background portion (bright portion) is green and the character portion (dark portion) is white, the vehicle is a commercial ordinary automobile. When the vehicle type is a light vehicle, the background portion (bright portion) is yellow and the character portion (dark portion) is black. In addition, when the background part (bright part) is black and the character part (dark part) is yellow, it is a commercial light vehicle. Moreover, the brightness | luminance of a specific part can be made into the average value of the brightness | luminance of a background part (bright part), for example.

  When the background part (bright part) of the license plate is set as the specific part, the average value of the luminance of the license plate area is obtained, and the area composed of pixels having a luminance higher than the average value is obtained as the specific part. it can.

  That is, the license plate detection unit 12 extracts, as a specific part, a bright area where the luminance value of each pixel in the detected license plate area is larger than the luminance threshold. The color of the background part excluding the character part such as the number (car number) is defined, for example, white (ordinary car etc.) or yellow (light car), and the luminance value is larger (high) than the character part Therefore, it is possible to extract a bright area having a luminance higher than a predetermined luminance threshold as a specific part. Since the color of a specific part is determined in advance, the brightness of the specific part (for example, an average value) is obtained, so that the color of the vehicle can be adjusted to sunlight conditions (lighting) such as the surrounding environment such as weather, time of day, sun or shade. It is possible to estimate how much it has been affected by the difference in (condition) (whether it has become brighter or darker).

  In addition, when the character part (dark part) of the license plate is used as the specific part, the average value of the brightness of the license plate area is obtained, and the area composed of pixels whose luminance is smaller than the average value is obtained as the specific part. Can do.

  That is, when the brightness value of the extracted bright area is equal to or higher than the predetermined upper limit value, the license plate detection unit 12 extracts a dark area where the brightness value of each pixel in the license plate area is smaller than the brightness threshold value as the specific part. For example, when the luminance value is represented by 0 to 255, the predetermined upper limit value can be a value of about 255 or 250 to 254. The dark area is, for example, a character portion of a license plate, and is green when the bright area is white, and black when the bright area is yellow. When the brightness value of the bright area is equal to or higher than the upper limit value, it can be determined that saturation (saturation) has occurred in the captured image. Therefore, the saturation is determined by setting the dark area as a specific part instead of the bright area. Even if a car has a problem, how much the color of the vehicle is affected by the difference in the sunshine conditions (lighting conditions) such as the weather, time of day, sun or shade, etc. ) Can be estimated.

  Note that the dark area is first extracted, and if the extracted dark area has a luminance value that is less than the predetermined lower limit value, the bright area where the luminance value of each pixel in the license plate area is greater than the predetermined luminance threshold is specified. It can also be extracted as a part.

  In addition, as a specific part, it is more preferable to use a bright part area | region rather than a dark part area | region. The reason is that the noise component generally included in the luminance value of the image follows a normal distribution having a constant variance regardless of the luminance value, so that the SN ratio is large in the bright part and small in the dark part. Therefore, the reliability is higher in the bright part than in the dark part. In addition, since the luminance values of the dark part and the bright part are calculated as average values, the larger the area (that is, the larger the number of pixels), the higher the statistical reliability. And generally, among vehicles traveling on the road, private cars with a large area of light areas are more common than commercial vehicles with a large area of dark areas, so it is more reliable to adopt light areas as a whole It is because it improves.

  The license plate color determination unit 14 determines the color of the pixel from the color component (for example, RGB value) of the pixel in the license plate area. For example, when the background portion of the license plate is a specific part, it can be determined from the color component of the pixel in the background portion of the license plate whether the pixel is white, yellow, green, or black. . When the character portion of the license plate is a specific part, it can be determined from the color component of the pixel of the character portion of the license plate whether the pixel is white, yellow, green, or black. .

  The vehicle body color determination area specifying unit 15 has a function as specifying means for specifying the determination area. The vehicle body color determination region specifying unit 15 specifies a determination region (vehicle body color determination region) for determining the color of the vehicle based on the position of the license plate region detected by the license plate detection unit 12 on the captured image.

  FIG. 3 is a schematic diagram illustrating an example of a vehicle body color determination region when the vehicle is imaged from the front. FIG. 4 is a schematic diagram illustrating an example of the vehicle body color determination region when the vehicle is imaged from the rear. As shown in FIG. 3, when imaging a vehicle from the front, a parallelogram-shaped area having a predetermined dimension along the road direction from the center of the upper edge of the detected license plate area (for example, a rectangular area). Can be used as a determination region. In the case of a vehicle head, the determination area is, for example, an area corresponding to a vehicle hood. In the example of FIG. 3, the determination region is specified such that the upper edge of the license plate region and the lower edge of the determination region are on the same line, but is not limited to the example of FIG. A determination region that is separated from the license plate region by an appropriate length may be specified.

  Also, as shown in FIG. 4, when the vehicle is imaged from the rear, the detected license plate area (for example, a rectangular area) is enclosed, and the license plate area is larger than the license plate area. The excluded area can be a determination area. In the case of a vehicle rear, the determination area is, for example, an area around the license plate excluding the rear window.

  The storage unit 21 stores a plurality of color sample tables in which a plurality of color components of pixels and the colors of the pixels are associated with each other corresponding to a plurality of luminance values.

  FIG. 5 is an explanatory diagram showing an example of the configuration of the color sample table according to the first embodiment. As shown in FIG. 5, a plurality of color sample tables are stored in association with a plurality of luminance values (luminances) Y1, Y2,. The plurality of luminance values are luminance values corresponding to respective brightnesses such as when the captured image is dark as a whole or when it is bright as a whole due to differences in sunshine conditions (illumination conditions). The number M of the plurality of luminance values can be set as appropriate. For example, eight luminance values can be set according to the degree of brightness.

  Each color sample table is a table in which a plurality of color components of a pixel are associated with the color of the pixel. The plurality of color components are, for example, R (red), G (green), and B (blue) values, but are not limited to RGB values, and values converted into another color system, for example, HIS values, YCbCr values, L * a * b * values (Elster, Aster, Beester), etc. can also be used.

  The pixel color is a determination color when determining the color of the vehicle, is a color that is desired to be classified when determining the vehicle body color, and the number of colors is N. The determination color is, for example, white, silver, gray, red, purple, yellow, blue, green, black, or the like. The determination color is not limited to this. The determination color can also include the color of the background portion (white, yellow) or character portion (green, black) of the license plate.

  FIG. 6 is an explanatory diagram showing the relationship between the color components (RGB values) of the color sample table and the vehicle body color. As shown in FIG. 6, when the color components are RGB, when each of RGB can take 256 values from 0 to 255, for example, the pixel color is determined in correspondence with all combinations of values. I can keep it. In other words, in the color sample table, any one of N determination colors is assigned to each of 256 cubes of RGB combinations. As shown in FIG. 5, the color component values (for example, 256 values for each of RGB) of the color sample table associated with each of the plurality of luminance values differ depending on the luminance value.

  The color sample table selection unit 16 has a function as a selection unit for selecting the color sample table, and selects a color sample table corresponding to the luminance value calculated by the license plate luminance value calculation unit 13. For example, when the background portion of the license plate is white, the color sample table corresponding to the average value of the luminance of the white portion (the luminance value matches or is closest) is selected. The color of the license plate is predetermined as white, so by calculating the luminance value (average luminance value), how bright or dark the body color is due to sunshine conditions (lighting conditions), etc. Can be estimated, and a color sample corresponding to the estimated brightness (average value of luminance) can be selected.

  The luminance value (average luminance, brightness) differs depending on whether the color of the pixel of the background portion or character portion of the license plate is white, yellow, green, or black, so the background portion of the license plate Alternatively, one luminance value (brightness) can be selected from a plurality of luminance values Y1,..., YM (brightness) based on the color of the pixel in the character portion and the average luminance of the color. Hereinafter, a method for selecting a luminance value (brightness) will be described.

  FIG. 7 is an explanatory diagram showing an example of a luminance selection table for selecting a color sample table. As shown in FIG. 7, the brightness selection table is divided into the bright part and dark part of the license plate, the average brightness value when the bright part color is white and yellow, and the average when the dark part color is green and black Brightness (brightness) Y1 to Y8 for selecting the color sample table is determined in association with the brightness value. The brightness selection table is for holding brightness information of the license plate when the brightness of the environment changes. In the case of a commercial ordinary vehicle or light vehicle, the only difference is that the bright and dark portions are reversed, so that the luminance can be selected in the same manner. Moreover, the numerical values in FIG. 7 are examples, and are not limited to these values.

  In the example of FIG. 7, for example, when the average luminance values of white, yellow, green, and black on the license plate are 80, 60, 20, and 10, respectively, the color table of luminance Y1 is selected. When the average luminance values of white, yellow, green, and black on the license plate are 100, 80, 30, and 20, respectively, the color table of luminance Y2 is selected. The same applies hereinafter. The color table selected according to the selected luminance values (Y1 to Y8) is as illustrated in FIG. In FIG. 7, a portion with a horizontal line instead of a number as the average luminance value represents a state in which the average luminance value is saturated to be greater than 250, for example.

  Then, as in Example 1 of FIG. 7, when the specific part is a bright part, the average luminance value of white (bright part) of the license plate is 100, and the average luminance value of green (dark part) is 60. In this case, the luminance Y2 corresponding to the average luminance value 100 of white (bright part) of the license plate is selected.

  Further, as in Example 2 in FIG. 7, when the specific portion is a bright portion, the average luminance value of white (bright portion) of the license plate is, for example, 255, and is a determination value as to whether or not it is saturated. When the average luminance value of green (dark part) is 100, the dark part is used as the specific part instead of the bright part. ) Is selected instead of the brightness Y6 corresponding to the average brightness value 100 of green (dark part).

  The vehicle body color determination unit 18 has a function as a determination unit that determines the color of the vehicle, and the color component and the selected color of each pixel in the determination region (vehicle body color determination region) specified by the vehicle body color determination region specifying unit 15 The color of the vehicle is determined based on the color components of the sample table.

  For example, a color corresponding to or close to a color component (for example, RGB value) of an arbitrary pixel in the determination region is determined as the color of the pixel, and the color is determined for all the pixels in the determination region. . If the number of pixels of the same color is equal to or greater than a predetermined threshold, the color of the pixel is determined as the color of the vehicle. As described above, by selecting a color sample table according to the luminance value of a specific part (for example, a background part, a bright part, etc.) of a license plate whose color is determined in advance, the influence of sunlight conditions (illumination conditions) etc. The color sample table can be used in advance, so accuracy is not affected by differences in sunlight conditions (lighting conditions) such as the weather at the time of imaging of the vehicle, the time zone, the sun or shade, and the surrounding environment. The color of the vehicle can be determined well.

  More specifically, the pixel number counting unit 17 serving as a counting unit counts, for each color, the number of pixels in which the color component of each pixel in the determination region (vehicle body color determination region) matches the color component of the color sample table. To do. When the number of pixels counted by the pixel number counting unit 17 is equal to or greater than the threshold, the vehicle body color determination unit 18 determines the color of the pixel as the color of the vehicle. If the color components are RGB, and each of RGB can take 256 values, for example, the color of the pixel can be determined in correspondence with all combinations of values. If the number of pixels having the same color component is determined and the number of pixels with the same color is equal to or greater than the threshold, the color component of the color sample table is determined by determining the color of the pixel as the vehicle color. Since the color of the vehicle can be determined only by referring to it, the determination process can be simplified.

  The edge pixel extraction unit 19 has a function as a pixel extraction unit, and extracts edge pixels based on the pixel value of each pixel in the determination region. For extraction of edge pixels, for example, an existing image filter such as a sobel filter can be used.

  The vehicle body color determination unit 18 determines the color of the vehicle by excluding (ignoring) edge pixels from the determination region. A vehicle bonnet or the like has almost no texture, whereas a light, emblem, or front grill has a texture. Further, a glass portion such as a window also has a texture due to light transmission or reflection. Regions with such textures do not accurately represent the color of the car body. By performing edge extraction and removing (ignoring) pixels with a strong edge strength (parts with texture) from the determination region, it is possible to prevent erroneous determination of the color of the vehicle. Note that the edge pixel extraction unit 19 is not an essential component, and may not be provided depending on the environmental situation or the like.

  The gradation area extraction unit 20 has a function as an area extraction unit, and extracts a gradation area having a uniform change in luminance value based on the luminance value of each pixel in the determination area.

  FIG. 8 is an explanatory diagram showing an example of a gradation area extraction method. FIG. 8A shows an example of the specified vehicle body color determination area (determination area). As shown in FIG. 8B, a linear gradation determination corresponding to the length in the vertical direction of the determination area along a predetermined direction in the determination area, in the example of FIG. 8B, along the vertical direction (Y-axis direction) of the captured image. Extract regions. Next, as shown in FIG. 8C, the gradation determination region is divided into small sections at a predetermined interval d, and the luminance value at each predetermined interval d is sampled. In the example of FIG. 8C, the luminance values I0, I1, I2, I3, I4, I5, and I6 are sampled.

  When there are luminance values that change constantly among the sampled luminance values, specifically, as shown in FIG. 8D, for example, (I2-I0), (I3-I1), (I4-I2) Are substantially equal and the absolute value of (I4-I0) is larger than a predetermined gradation threshold, that is, the difference in luminance value (for example, I2-I0) at an appropriate interval (2d in FIG. 8) is substantially constant. If the absolute value of the change amount of the luminance value (I4-I0 in FIG. 8) satisfies the condition larger than the gradation threshold, the region including the range of the length L obtained by sampling the luminance values I0 to I4 is defined as the gradation region. Can be extracted as Note that the sampled luminance values I5 and I6 do not satisfy the above-described conditions.

  FIG. 9 is an explanatory diagram showing a state in which the gradation area is excluded from the determination area. Since the area of length L in the vertical direction is extracted as a gradation area from FIG. 8, the gradation area of length L in the vertical direction is excluded from the original vehicle body color determination area (determination area) as shown in FIG. The region is a vehicle body color determination region. On the surface facing up, such as a bonnet, the color of the sky may appear due to the reflection of light, appearing as a gradation, and the color of the vehicle may appear different. By removing (ignoring) the gradation area from the determination area, it is possible to prevent the vehicle color from being erroneously determined.

  In the example of FIG. 8, the gradation determination area is extracted along the vertical direction (Y-axis direction). However, the present invention is not limited to this, and the gradation determination area is extracted along the horizontal direction (X-axis direction). You can also In the example of FIG. 8, it is determined whether one gradation determination area is extracted and the above-described condition is satisfied. It can also be determined whether or not the condition is satisfied. Further, the gradation area extraction unit 20 is not an essential configuration and may not be provided depending on the environmental situation or the like.

  Next, the operation of the vehicle color determination device 100 will be described. FIG. 10 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device 100 according to the first embodiment. Hereinafter, the main body of the processing will be described as the control unit 10. The control unit 10 acquires a captured image (S11), and detects a license plate area in the acquired captured image (S12).

  The control unit 10 specifies the vehicle body color determination region (determination region) based on the detected position of the license plate region on the captured image (S13). The control unit 10 calculates a luminance value (for example, an average value) of a specific part (for example, a background portion or a bright portion) of the detected license plate region (S14). The specific part can be specified based on the luminance or the color component.

  The control unit 10 selects a color sample table of luminance (for example, Y1,... YM in FIG. 5) corresponding to the calculated luminance value from the stored color sample table (S15). The luminance (for example, Y1,... YM in FIG. 5) corresponding to the calculated luminance value means, for example, the luminance (Y1,... YM) or the closest luminance (Y1,... YM) that matches the calculated luminance value. To do.

  The control unit 10 counts, for each color, the number of pixels whose color components match the color components of the selected color sample table for each pixel in the vehicle body color determination area (S16). Thereby, it is determined whether each pixel in the vehicle body color determination area is a predetermined determination color (for example, white, silver, gray, red, purple, yellow, blue, green, black, etc.). At the same time, the number of pixels determined to be the determination color is counted for each determination color.

  The control unit 10 determines whether or not the counted number of pixels is equal to or greater than the threshold (S17). If the number of pixels is equal to or greater than the threshold (YES in S17), the control unit 10 determines that the color of the pixel is the body color. (S18), the process ends. If the number of pixels is not equal to or greater than the threshold (NO in S17), the control unit 10 ends the process without performing the process of step S18. The process shown in FIG. 10 is repeatedly performed during the determination of the vehicle body color.

(Embodiment 2)
In the first embodiment described above, the RGB values are used as the color components of the pixels in the color sample table. However, the color components are not limited to the RGB values, and other color systems such as, for example, , L * a * b * values (Elster, Aster, Beester) can also be used.

  FIG. 11 is a block diagram illustrating an example of the configuration of the vehicle color determination device 120 according to the second embodiment. The difference from the first embodiment is that an L * a * b * value calculation unit 22 and an approximation degree calculation unit 23 are provided. The L * a * b * value can be converted from the RGB value by a known conversion formula.

  The storage unit 21 stores in advance L * a * b * values corresponding to each color (determination color for determining the color of the vehicle) in the color sample table.

  FIG. 12 is an explanatory diagram showing an example of the configuration of the color sample table according to the second embodiment. As shown in FIG. 12, a plurality of color sample tables are stored in association with a plurality of luminance values (luminances) Y1, Y2,. The plurality of luminance values are luminance values corresponding to respective brightnesses such as when the captured image is dark as a whole or when it is bright as a whole due to differences in sunshine conditions (illumination conditions). The number M of the plurality of luminance values can be set as appropriate. For example, eight luminance values can be set according to the degree of brightness.

  Each color sample table is a table in which a plurality of color components of a pixel are associated with the color of the pixel. The plurality of color components are L * a * b * values (Elster, Aster, Biester). The L * a * b * color system is a color system designed to express the difference in color perceived by humans with numerical values, and outputs results that are closer to the human sense than other color systems. Is possible.

  The pixel color is a determination color when determining the color of the vehicle, is a color that is desired to be classified when determining the vehicle body color, and the number of colors is N. The determination color is, for example, white, silver, gray, red, purple, yellow, blue, green, black, or the like. The determination color is not limited to this. The determination color can also include the color of the background portion (white, yellow) or character portion (green, black) of the license plate.

  FIG. 13 is an explanatory diagram showing the relationship between the color components (L * a * b * values) of the color sample table and the body color. As shown in FIG. 13, when the color component is L * a * b *, each L * a * b * value can be used as a reference value, and the pixel color can be determined for each reference value. That is, the color sample table associates one L * a * b * value as a reference value with each of the N determination colors. As shown in FIG. 12, the color component value (L * a * b * value) of the color sample table associated with each of the plurality of luminance values varies depending on the luminance value.

  The color sample table selection unit 16 has a function as a selection unit for selecting the color sample table, and selects a color sample table corresponding to the luminance value calculated by the license plate luminance value calculation unit 13. For example, when the background portion of the license plate is white, the color sample table corresponding to the average value of the luminance of the white portion (the luminance value matches or is closest) is selected. The color of the license plate is predetermined as white, so by calculating the luminance value (average luminance value), how bright or dark the body color is due to sunshine conditions (lighting conditions), etc. Can be estimated, and a color sample corresponding to the estimated brightness (average value of luminance) can be selected.

  The L * a * b * value calculation unit 22 calculates the L * a * b * value of each pixel based on the color component (for example, RGB value) of each pixel in the determination area. The L * a * b * value can be obtained from the RGB value using a predetermined conversion formula, for example.

  The approximation calculation unit 23 calculates the approximation between the L * a * b * value calculated by the L * a * b * value calculation unit 22 and the L * a * b * value of each color in the color sample table. Calculate for pixels. The degree of approximation is an index indicating how close the L * a * b * value of the pixel in the determination area and the L * a * b * value of the color sample table are, and each L * a * b * When the sum of the squares of the differences is the smallest, it can be assumed that the degree of approximation is the largest.

  The pixel number counting unit 17 counts the number of pixels whose approximation degree calculated by the approximation degree calculation unit 23 is equal to or greater than a predetermined value.

  When the number of pixels counted by the pixel number counting unit 17 is equal to or greater than the threshold, the vehicle body color determination unit 18 determines the color of the pixel as the color of the vehicle.

  FIG. 14 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device 120 according to the second embodiment. Hereinafter, the main body of the processing will be described as the control unit 10. The control unit 10 acquires a captured image (S31), and detects a license plate area in the acquired captured image (S32).

  The control unit 10 specifies a vehicle body color determination area (determination area) based on the detected position of the license plate area on the captured image (S33). The control unit 10 calculates a luminance value (for example, an average value) of a specific part (for example, a background portion or a bright portion) of the detected license plate region (S34). The specific part can be specified based on the luminance or the color component.

  The control unit 10 selects a color sample table of luminance (for example, Y1,... YM in FIG. 12) corresponding to the calculated luminance value from the stored color sample table (S35). The luminance corresponding to the calculated luminance value (for example, Y1,... YM in FIG. 12) means, for example, the luminance (Y1,... YM) that matches the calculated luminance value or the closest luminance (Y1,... YM). To do.

  The control unit 10 converts the color component (for example, RGB value) of each pixel in the vehicle body color determination region into an L * a * b * value (S36), and L * a * b * of each pixel in the vehicle body color determination region. The degree of approximation between the value and each L * a * b * value of the color sample table is calculated (S37).

  The control unit 10 counts the number of pixels whose calculated degree of approximation is a predetermined value or more (S38). Thereby, it is determined whether each pixel in the vehicle body color determination area is a predetermined determination color (for example, white, silver, gray, red, purple, yellow, blue, green, black, etc.). At the same time, the number of pixels determined to be the determination color is counted for each determination color.

  The control unit 10 determines whether or not the counted number of pixels is equal to or greater than the threshold (S39). If the number of pixels is equal to or greater than the threshold (YES in S39), the control unit 10 determines that the color of the pixel is the body color. (S40), the process ends. If the number of pixels is not greater than or equal to the threshold (NO in S39), the control unit 10 ends the process without performing the process of step S40. The process shown in FIG. 14 is repeatedly performed during the determination of the vehicle body color.

  In the second embodiment described above, the RGB value is converted into the Lab value for each pixel, the degree of approximation with the Lab value of the color sample is calculated, and the color is determined. However, the conversion process is performed for each pixel. When processing such as calculation of the degree of approximation is performed, the calculation load increases. Therefore, all RGB combinations (256 to the cube of 256) are converted into Lab values in advance, approximate values of Lab swatches with color samples are calculated, colors are determined, and all RGB combinations (256 By preparing a correspondence table of colors determined to be (the third power of 3), when the RGB value of a pixel is obtained, the color can be determined immediately without performing a calculation by referring to the table. . The table is configured as shown in FIG.

  By using such a table, there is always a corresponding RGB combination in the table even if it is an arbitrary RGB value, so an intermediate calculation for determining the color (for example, calculation of the approximation degree) is performed. It becomes unnecessary. Further, by treating the input 256 gradations as, for example, 32 gradations, the size (size) of the table can be reduced, and both reduction in calculation load and reduction in memory capacity can be achieved.

(Embodiment 3)
In the first and second embodiments described above, a plurality of color sample tables are stored according to the luminance value, and the color sample table corresponding to the luminance value is selected according to the luminance value of the license plate area. However, the present invention is not limited to this. For example, it is possible to store one color sample table corresponding to a general or representative luminance value and determine the vehicle body color with reference to the color sample table.

  FIG. 15 is a block diagram illustrating an example of the configuration of the vehicle color determination device 140 according to the third embodiment. The difference from the first embodiment is that a luminance value correction unit 24 is provided.

  The storage unit 21 stores one color sample table in which a plurality of color components of a pixel and the color of the pixel are associated with each other as illustrated in FIG. The predetermined luminance value is, for example, a luminance value corresponding to average or general brightness when the luminance of the entire captured image changes due to a difference in sunshine conditions (illumination conditions). The brightness value can be set. The plurality of color components are, for example, R (red), G (green), and B (blue) values. The pixel color is a determination color when determining the color of the vehicle, and is, for example, white, silver, gray, red, purple, yellow, blue, green, black, or the like. The determination color can also include the color of the background portion (white, yellow) or character portion (green, black) of the license plate. If the color components are RGB, and each of RGB can take 256 values, for example, the color of the pixel can be determined in correspondence with all combinations of values.

  The luminance value correction unit 24 has a function as a correction unit, and the luminance value of each pixel in the determination region according to the difference between the luminance value calculated by the license plate luminance value calculation unit 13 and the predetermined luminance value described above. Correct. For example, the luminance value of the entire determination area is corrected so that the calculated luminance value matches a predetermined luminance value.

  FIG. 16 is a schematic diagram illustrating an example of luminance value correction according to the third embodiment. As shown in FIG. 16, the average brightness of the detected license plate area (for example, the background portion and the bright portion) is compared with the average brightness (predetermined brightness value) of the color sample table, and the two brightnesses match or The luminance value of the entire captured image (or the entire determination area) may be corrected so as to approximate it. For example, when the calculated luminance value is smaller than a predetermined luminance value, that is, when the average value of the luminance of a specific part of the license plate area is smaller than the predetermined luminance value, the entire captured image is displayed as shown in FIG. Since the brightness of the determination area is increased, the brightness value is corrected so that the brightness of the determination area matches or approximates the brightness of the color sample table.

  Further, when the calculated luminance value is larger than the predetermined luminance value, that is, when the average luminance value of the specific part of the license plate area is larger than the predetermined luminance value, the captured image is bright as a whole, so the determination area And the luminance value is corrected so that the luminance of the determination area matches or approximates the luminance of the color sample table. As a result, the luminance of the determination region can be matched or brought close to the luminance of the color sample.

  The vehicle body color determination unit 18 determines the color of the vehicle based on the color component of each pixel in the corrected determination area and the color component of the color sample table. For example, a color corresponding to or close to a color component (for example, RGB value) of an arbitrary pixel in the determination region is determined as the color of the pixel, and the color is determined for all the pixels in the determination region. . If the number of pixels of the same color is equal to or greater than a predetermined threshold, the color of the pixel is determined as the color of the vehicle.

  In Embodiment 3, since the luminance value of the determination region is corrected so that the luminance value of a specific part of the license plate whose color is determined in advance matches (or approaches) the predetermined luminance value, the sunshine condition (illumination condition) The brightness value of the judgment area can be matched or brought close to the brightness of the color sample by offsetting the influence of the lighting conditions, etc., and the sunshine conditions (lighting conditions) such as the weather at the time of imaging of the vehicle, the time zone, the sun or shade The color of the vehicle can be accurately determined without being affected by the difference in

  FIG. 17 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device 140 according to the third embodiment. Hereinafter, the main body of the processing will be described as the control unit 10. The control unit 10 acquires a captured image (S51), and detects a license plate area in the acquired captured image (S52).

  The control unit 10 specifies the vehicle body color determination region (determination region) based on the detected position of the license plate region on the captured image (S53). The control unit 10 calculates a luminance value (for example, an average value) of a specific part (for example, a background portion or a bright portion) of the detected license plate region (S54). The specific part can be specified based on the luminance or the color component.

  The control unit 10 corrects the luminance value of each pixel in the vehicle body color determination area according to the difference between the calculated luminance value and a predetermined luminance value (luminance value of the color sample table) (S55).

  The control unit 10 counts the number of pixels whose color components match the color components of the color sample table for each pixel of the corrected vehicle body color determination region (S56). Thereby, it is determined whether each pixel in the vehicle body color determination area is a predetermined determination color (for example, white, silver, gray, red, purple, yellow, blue, green, black, etc.). At the same time, the number of pixels determined to be the determination color is counted for each determination color.

  The control unit 10 determines whether or not the counted number of pixels is equal to or greater than the threshold (S57). If the number of pixels is equal to or greater than the threshold (YES in S57), the control unit 10 determines that the color of the pixel is the body color. (S58), the process ends. If the number of pixels is not greater than or equal to the threshold value (NO in S57), the control unit 10 ends the process without performing the process of step S58. Note that the process shown in FIG. 17 is repeatedly performed during the determination of the vehicle body color.

(Embodiment 4)
In Embodiment 3 described above, the RGB values are used as the color components of the pixels in the color sample table. However, the color components are not limited to the RGB values, and other color systems such as, for example, , L * a * b * values (Elster, Aster, Beester) can also be used.

  FIG. 18 is a block diagram illustrating an example of the configuration of the vehicle color determination device 160 according to the fourth embodiment. The difference from Embodiment 3 is that an L * a * b * value calculation unit 22 and an approximation degree calculation unit 23 are provided. The L * a * b * value can be converted from the RGB value by a known conversion formula.

  The storage unit 21 stores one color sample table in which a plurality of color components of a pixel and the color of the pixel are associated with each other as shown in FIG. The predetermined luminance value is, for example, a luminance value corresponding to average or general brightness when the luminance of the entire captured image changes due to a difference in sunshine conditions (illumination conditions). The brightness value can be set. The plurality of color components are, for example, L * a * b * values. The pixel color is a determination color when determining the color of the vehicle, and is, for example, white, silver, gray, red, purple, yellow, blue, green, black, or the like. The determination color can also include the color of the background portion (white, yellow) or character portion (green, black) of the license plate. When the color component is L * a * b *, each L * a * b * value can be used as a reference value, and the pixel color can be determined for each reference value. That is, the color sample table associates one L * a * b * value as a reference value with each of the N determination colors.

  The luminance value correction unit 24 has a function as a correction unit, and the luminance value of each pixel in the determination region according to the difference between the luminance value calculated by the license plate luminance value calculation unit 13 and the predetermined luminance value described above. Correct. For example, the luminance value of the entire determination area is corrected so that the calculated luminance value matches a predetermined luminance value. The correction of the luminance value is the same as in the example of FIG.

  The L * a * b * value calculation unit 22 calculates the L * a * b * value of each pixel based on the color component (for example, RGB value) of each pixel in the determination region in which the luminance value is corrected. The L * a * b * value can be obtained from the RGB value using a predetermined conversion formula, for example.

  The approximation calculation unit 23 calculates the approximation between the L * a * b * value calculated by the L * a * b * value calculation unit 22 and the L * a * b * value of each color in the color sample table. Calculate for pixels. The degree of approximation is an index indicating how close the L * a * b * value of the pixel in the determination area and the L * a * b * value of the color sample table are, and each L * a * b * When the sum of the squares of the differences is the smallest, it can be assumed that the degree of approximation is the largest.

  The pixel number counting unit 17 counts the number of pixels whose approximation degree calculated by the approximation degree calculation unit 23 is equal to or greater than a predetermined value.

  When the number of pixels counted by the pixel number counting unit 17 is equal to or greater than the threshold, the vehicle body color determination unit 18 determines the color of the pixel as the color of the vehicle.

  In the fourth embodiment, the luminance value of the determination region is corrected so that the luminance value of the specific part of the license plate whose color is determined in advance matches (or approaches) the predetermined luminance value. Therefore, the sunshine condition (illumination condition) The brightness value of the judgment area can be matched or brought close to the brightness of the color sample by offsetting the influence of the lighting conditions, etc., and the sunshine conditions (lighting conditions) such as the weather at the time of imaging of the vehicle, the time zone, the sun or shade The color of the vehicle can be accurately determined without being affected by the difference in

  FIG. 19 is a flowchart illustrating an example of a processing procedure of the vehicle color determination device 160 according to the fourth embodiment. Hereinafter, the main body of the processing will be described as the control unit 10. The control unit 10 acquires a captured image (S71), and detects a license plate area in the acquired captured image (S72).

  The control unit 10 specifies a vehicle body color determination region (determination region) based on the detected position of the license plate region on the captured image (S73). The control unit 10 calculates a luminance value (for example, an average value) of a specific part (for example, a background portion or a bright portion) of the detected license plate region (S74). The specific part can be specified based on the luminance or the color component.

  The control unit 10 corrects the luminance value of each pixel in the vehicle body color determination area according to the difference between the calculated luminance value and a predetermined luminance value (luminance value of the color sample table) (S75).

  The control unit 10 converts the color component (for example, RGB value) of each pixel in the vehicle body color determination region into an L * a * b * value (S76), and L * a * b * of each pixel in the vehicle body color determination region. The degree of approximation between the value and each L * a * b * value of the color sample table is calculated (S77).

  The control unit 10 counts the number of pixels whose calculated degree of approximation is a predetermined value or more (S78). Thereby, it is determined whether each pixel in the vehicle body color determination area is a predetermined determination color (for example, white, silver, gray, red, purple, yellow, blue, green, black, etc.). At the same time, the number of pixels determined to be the determination color is counted for each determination color.

  The control unit 10 determines whether or not the counted number of pixels is equal to or greater than the threshold (S79). If the number of pixels is equal to or greater than the threshold (YES in S79), the control unit 10 determines that the color of the pixel is the body color. (S80), the process ends. If the number of pixels is not greater than or equal to the threshold (NO in S79), the control unit 10 ends the process without performing the process of step S80. Note that the process shown in FIG. 19 is repeatedly performed during the determination of the vehicle body color.

  The vehicle color determination devices 100, 120, 140, and 160 described above can be realized using a general-purpose computer that includes a CPU, a RAM, and the like. That is, as shown in FIG. 10, FIG. 14, FIG. 17, and FIG. 19, a computer program that defines each processing procedure is recorded on a recording medium, and the computer program is loaded from the recording medium into a RAM provided in the computer. By executing the computer program on the CPU, the vehicle color determination device can be realized on the computer.

  In the above-described first to fourth embodiments, when the number of pixels counted by the pixel number counting unit is equal to or greater than the threshold, the color of the pixel is determined as the vehicle body color. However, the present invention is not limited to this. The color of the pixel with the largest number of counted pixels can be determined as the vehicle body color.

  In the above-described example, the color of the pixel in the specific part is determined based on the color component (for example, RGB value) of the pixel in the license plate region. However, the present invention is not limited to this. The color of the pixel at the specific part can also be determined by this method.

  The above-described vehicle color determination devices 100, 120, 140, and 160 of the present embodiment also have a function as a license plate color determination device. The license plate color determination apparatus according to the present embodiment includes configurations such as a license plate detection unit 12, a license plate luminance value calculation unit 13, a license plate color determination unit 14, and the like. In the following method, it can also be determined whether or not the vehicle is a light vehicle.

  The license plate detection unit 12 extracts a captured image from the storage unit 21 and detects a vehicle (including a part of the vehicle) based on the extracted captured image. And the license plate detection part 12 detects a license plate area | region by performing the cutting-out process of a rectangular license plate part based on the detected vehicle image. The license plate area can be specified by coordinates on the captured image of the four corners of the rectangle.

  The license plate luminance value calculation unit 13 calculates the luminance Y of each pixel in the license plate region detected by the license plate detection unit 12.

  The license plate color determination unit 14 calculates the color differences Cb and Cr of each pixel in the license plate area detected by the license plate detection unit 12.

  The luminance Y can be calculated by the equation Y = 0.298912 × R + 0.586611 × G + 0.114477 × B. Further, the color difference Cb can be calculated by an equation of Cb = −0.168777 × R−0.331223 × G + 0.5 × B. Further, the color difference Cr can be calculated by an equation of Cr = 0.5 × R−0.418358 × G−0.081642 × B. The range of each of RGB is, for example, 0 to 255, the range of luminance Y is 0 to 255, and the range of color differences Cb and Cr is −128 to 127.

  The license plate color determination unit 14 specifies a determination region of a vehicle, for example, a light vehicle, based on the license plate region detected by the license plate detection unit 12. The determination area of a light vehicle can be, for example, an area inside the license plate area and having a shape similar to the license plate area. The detected license plate region may include a portion other than the actual license plate due to an error during detection. And parts other than the license plate become noise and may reduce the accuracy of the color determination process. By setting the determination area of the light vehicle to the inside of the detected license plate area, it is possible to prevent a portion other than the actual license plate from being included, and to increase the determination accuracy.

  The license plate color determination unit 14 calculates an average value Ya of the luminance values in the entire determination region, and identifies an area composed of pixels having a luminance value equal to or higher than the average value Ya as a bright portion (specific part). Then, an area composed of pixels whose luminance value is smaller than the average value Ya is specified as a dark part (specific part). That is, the license plate color determination unit 14 specifies a specific part according to the brightness of the determination region.

  The license plate color determination unit 14 calculates the average value Ya of the brightness of the bright part, the average value Cba of the color difference Cb, and the average value Cra of the color difference Cr, and the number based on the calculated average value of the color difference and a predetermined color difference threshold value. Determine the color of the plate. Here, the color of the license plate means that when the bright part is a specific part, it includes the color of the part other than the bright part and the bright part of the license plate area (determination area) and the dark part is the specific part. Includes the colors of the dark part and the part other than the dark part of the license plate area (determination area). The license plate color determination unit 14 determines whether the vehicle is a light vehicle based on the calculated average value of color differences and a predetermined color difference threshold.

  Specific determination conditions can be as follows, for example. That is, as the condition 1, when the average value Cba of the color difference Cb is smaller than the color difference threshold value Tb (Cba <Tb), it is assumed that the color of the bright portion is yellow and the background portion of the license plate is yellow, so the vehicle is a light vehicle. judge.

  When the range of the color difference Cb is −128 to 127, positive represents a blue component and negative represents a yellow component. That is, by setting the color difference threshold Tb to a value such that the color difference Cb represents a yellow component, when the average value Cba of the color difference Cb is smaller than the predetermined color difference threshold Tb (for example, the average value Cba is negative). Value), it can be determined that the color of the bright portion is yellow.

  In addition, when the average value Cba of the color difference Cb is larger than the predetermined color difference threshold Tb (for example, the average value Cba is a positive value), since a blue component closer to white than yellow is included, the light color is white. It can be determined that

  As described above, it is possible to determine the color of the license plate simply by comparing the color difference and the color difference threshold. Compared to the conventional case where the values of RGB and the thresholds corresponding to RGB are compared, The license plate color determination process can be simplified.

  Further, as the condition 2, when the average value Cra of the color difference Cr is larger than the color difference threshold Tr (Cra> Tr), it is assumed that the color of the bright part is yellow, and the background portion of the license plate is yellow, so the vehicle is a light vehicle. judge. Instead of the average value of luminance and color difference, a median value between the maximum value and the minimum value can also be used.

  When the range of the color difference Cr is −128 to 127, positive represents a red component and negative represents a light blue component. That is, by setting the color difference threshold Tr to a value such that the color difference Cr represents a red component, when the average value Cra of the color difference Cr is larger than the predetermined color difference threshold Tr (for example, the average value Cra is positive). Value), since a red component closer to yellow than white is included, it can be determined that the color of the bright portion is yellow close to red.

  In addition, when the average value Cra of the color difference Cr is smaller than the predetermined color difference threshold Tr (for example, the average value Cra is a negative value), since a light blue component closer to white than yellow is included, the light color is white. It can be determined that

  As described above, it is possible to determine the color of the license plate simply by comparing the color difference and the color difference threshold. The license plate color determination process can be simplified.

  Further, as condition 3, when both conditions 1 and 2 are satisfied, it can be determined that the light portion is yellow, and the background portion of the license plate is yellow, so that it is a light vehicle.

  When RGB values are used for light vehicle determination, a comparison determination process with a threshold value is required for each of the RGB values. By using the color differences Cb and Cr instead of the RGB values, it is only necessary to compare and determine the color difference Cb or Cr value with one threshold value, so that the processing can be simplified.

  FIG. 20 is a flowchart illustrating a light vehicle determination processing procedure performed by the vehicle color determination devices according to the first to fourth embodiments. Note that the process illustrated in FIG. 20 shows an example of a license plate color determination method by the license plate color determination apparatus of the present embodiment.

  The control unit 10 acquires a captured image (S101), and detects a license plate area in the acquired captured image (S102).

  The control unit 10 specifies the determination area of the light vehicle on the basis of the detected position of the license plate area on the captured image (S103). The determination area of the light vehicle is, for example, an area inside the license plate area and having a similar shape to the license plate area. The control unit 10 calculates the luminance Y and color differences Cb and Cr of each pixel in the determination area of the light vehicle (S104).

  The control unit 10 calculates the average value of the luminance in the determination area of the light vehicle (S105), and identifies the bright part in the determination area based on the calculated average value (S106).

  The control unit 10 calculates the average brightness Ya and average color differences Cba and Cra of the specified bright part (S107). The control unit 10 determines whether or not the average color difference Cba is smaller than the color difference threshold Tb (S108). If the average color difference Cba is smaller than the color difference threshold Tb (YES in S108), the light portion is assumed to be yellow, and the light vehicle (S109) and the process is terminated.

  When the average color difference Cba is not smaller than the color difference threshold Tb (NO in S108), the control unit 10 determines whether the average color difference Cra is larger than the color difference threshold Tr (S110), and the average color difference Cra is larger than the color difference threshold Tr. If so (YES in S110), the process of step S109 is performed. When the average color difference Cra is not larger than the color difference threshold Tr (NO in S110), the control unit 10 determines that the bright part is not yellow, determines that the light part is not a light vehicle (S111), and ends the process. When it is determined that the vehicle is a light vehicle, it is determined that the bright portion is yellow and the dark portion is black. When it is determined that the vehicle is not a light vehicle, the bright portion is determined to be white and the dark portion is determined to be green.

  In the process of FIG. 20 described above, one of steps S108 and S110 may be omitted. Further, when the average color difference Cba is smaller than the color difference threshold Tb and the average color difference Cra is larger than the color difference threshold Tr, it may be determined that the vehicle is a light vehicle.

  The above-described license plate color determination device can also be realized using a general-purpose computer including a CPU, a RAM, and the like. That is, as shown in FIG. 20, a computer program that defines each processing procedure is recorded on a recording medium, the computer program is loaded from the recording medium into a RAM provided in the computer, and the computer program is executed by the CPU. Thus, a license plate color determination device can be realized on a computer.

  The disclosed embodiments are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

DESCRIPTION OF SYMBOLS 100, 120, 140, 160 Car color determination apparatus 10 Control part 11 Interface part 12 License plate detection part 13 License plate luminance value calculation part 14 License plate color determination part 15 Car body color determination area specification part 16 Color sample table selection part 17 Pixel Number counting unit 18 Car body color determining unit 19 Edge pixel extracting unit 20 Gradation region extracting unit 21 Storage unit 22 L * a * b * value calculating unit 23 Approximation degree calculating unit 24 Luminance value correcting unit 200 Imaging device

Claims (12)

In a vehicle color determination device that determines the color of a vehicle based on a captured image obtained by imaging the vehicle,
Detecting means for detecting a license plate area based on a pixel value of each pixel constituting the captured image;
Luminance value calculating means for calculating the luminance value of a specific part of the license plate area detected by the detecting means;
Storage means for storing a plurality of color sample tables in which a plurality of color components of pixels and the colors of the pixels are associated with each other in correspondence with a plurality of luminance values;
Selecting means for selecting a color sample table corresponding to the brightness value calculated by the brightness value calculating means;
Specifying means for specifying a determination area for determining the color of the vehicle based on the position on the captured image of the license plate area detected by the detection means;
Vehicle color determination, comprising: a determination unit that determines a color of the vehicle based on a color component of each pixel in the determination region specified by the specifying unit and a color component of a color sample table selected by the selection unit apparatus. A counting means for counting the number of pixels in which the color component of each pixel in the determination area matches the color component of the color sample table for each color;
The determination means includes
2. The vehicle color determination device according to claim 1, wherein when the number of pixels counted by the counting means is equal to or greater than a threshold, the color of the pixel is determined as the color of the vehicle. The storage means
L * a * b * values corresponding to the respective colors in the color sample table are stored,
Calculating means for calculating an L * a * b * value of each pixel based on a color component of each pixel in the determination region;
Approximation degree calculating means for calculating an approximation degree between the L * a * b * value calculated by the calculating means and the L * a * b * value of each color of the color sample table for each pixel in the determination region;
Counting means for counting the number of pixels whose approximation degree calculated by the approximation degree calculation means is a predetermined value or more,
The determination means includes
2. The vehicle color determination device according to claim 1, wherein when the number of pixels counted by the counting means is equal to or greater than a threshold, the color of the pixel is determined as the color of the vehicle.   4. The apparatus according to claim 1, further comprising: an extraction unit that extracts a bright area where the luminance value of each pixel of the license plate area detected by the detection unit is larger than a luminance threshold as the specific part. The vehicle color determination device according to item. The extraction means includes
When the brightness value of the bright area is equal to or higher than a predetermined upper limit value, a dark area where the brightness value of each pixel of the license plate area is smaller than the brightness threshold is extracted as the specific part. The vehicle color determination device according to claim 4, wherein Comprising pixel extraction means for extracting an edge pixel based on the pixel value of each pixel in the determination region;
The specifying means is:
6. The vehicle color determination device according to claim 1, wherein the edge pixel extracted by the pixel extraction means is excluded from the specified determination region. A region extracting means for extracting a gradation region in which a change in luminance value is uniform based on a luminance value of each pixel of the determination region;
The specifying means is:
The vehicle color determination device according to any one of claims 1 to 6, wherein the gradation region extracted by the region extraction unit is excluded from the specified determination region. In a vehicle color determination device that determines the color of a vehicle based on a captured image obtained by imaging the vehicle,
Detecting means for detecting a license plate area based on a pixel value of each pixel constituting the captured image;
Luminance value calculating means for calculating the luminance value of a specific part of the license plate area detected by the detecting means;
Storage means for storing a color sample table in which a plurality of color components of a pixel and the color of the pixel are associated with each other in correspondence with a predetermined luminance value;
Specifying means for specifying a determination area for determining the color of the vehicle based on the position on the captured image of the license plate area detected by the detection means;
Correction means for correcting the luminance value of each pixel in the determination area according to the difference between the luminance value calculated by the luminance value calculating means and the predetermined luminance value;
A vehicle color determination apparatus comprising: determination means for determining a color of a vehicle based on a color component of each pixel in the determination region corrected by the correction means and a color component of the color sample table. In a computer program for causing a computer to determine the color of a vehicle based on a captured image obtained by imaging the vehicle,
On the computer,
Detecting a license plate region based on a pixel value of each pixel constituting the captured image;
Calculating a luminance value of a specific part of the detected license plate region;
Selecting a color sample table corresponding to the calculated luminance value from a plurality of color sample tables in which a plurality of color components of the pixel and the color of the pixel are associated with each other corresponding to the plurality of luminance values;
Identifying a determination area for determining the color of the vehicle based on the position of the detected license plate area on the captured image;
A computer program for executing a step of determining a color of a vehicle based on a color component of each pixel in a specified determination region and a color component of a selected color sample table. In a computer program for causing a computer to determine the color of a vehicle based on a captured image obtained by imaging the vehicle,
On the computer,
Detecting a license plate region based on a pixel value of each pixel constituting the captured image;
Calculating a luminance value of a specific part of the detected license plate region;
Identifying a determination area for determining the color of the vehicle based on the position of the detected license plate area on the captured image;
Correcting the luminance value of each pixel in the determination region according to the difference between the calculated luminance value and a predetermined luminance value;
The vehicle color is determined based on the color component of each pixel in the corrected determination region, the color component of the pixel corresponding to the predetermined luminance value, and the color component of the color sample table in which the color of the pixel is associated. A computer program for executing the steps. In the vehicle color determination method by the vehicle color determination device that determines the color of the vehicle based on the captured image obtained by imaging the vehicle,
A plurality of color sample tables in which a plurality of color components of a pixel and the color of the pixel are associated with each other corresponding to a plurality of luminance values are stored,
Detecting a license plate region based on a pixel value of each pixel constituting the captured image;
Calculating a luminance value of a specific part of the detected license plate region;
Selecting a color sample table corresponding to the calculated luminance value;
Identifying a determination region for determining the color of the vehicle based on the position on the captured image of the detected license plate region;
Determining the color of the vehicle based on the color component of each pixel in the specified determination region and the color component of the selected color sample table. In the vehicle color determination method by the vehicle color determination device that determines the color of the vehicle based on the captured image obtained by imaging the vehicle,
A color sample table in which a plurality of color components of a pixel and the color of the pixel are associated with each other corresponding to a predetermined luminance value is stored.
Detecting a license plate region based on a pixel value of each pixel constituting the captured image;
Calculating a luminance value of a specific part of the detected license plate region;
Identifying a determination region for determining the color of the vehicle based on the position on the captured image of the detected license plate region;
Correcting the luminance value of each pixel in the determination region according to the difference between the calculated luminance value and the predetermined luminance value;
A vehicle color determination method comprising: determining a color of a vehicle based on a color component of each pixel in the corrected determination region and a color component of the color sample table.

Images