JP2007163173A - Apparatus, method, and program for measuring vehicle - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus for measuring a vehicle by a simple configuration capable of highly accurately measuring the outside dimensions of vehicles and to provide a method of measuring the vehicle and a program for measuring the vehicle. <P>SOLUTION: When the outside dimensions of the vehicle are to be measured on the basis of images, two-dimensional images are generated by executing the processing of photographing a prescribed range in a direction perpendicular to the traveling direction of the vehicle traveling over a road surface and acquiring images of the vehicle traversing in front of a background part a plurality of times. A reference image is specified on the basis of parts of the two-dimensional images which do not include the vehicle. By sequentially comparing the reference image with other images, images having the differences between them exceeding a threshold are extracted as images of the vehicle. The outside dimensions of the vehicle are measured on the basis of the extracted images. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a vehicle measurement device, a vehicle measurement method, and a vehicle measurement program for measuring an outer dimension of a vehicle.

As a technique for measuring the outer dimension of a vehicle, a vehicle measuring device that captures a vehicle and measures the outer dimension of the vehicle from the image is known. For example, in Patent Document 1, a vehicle traveling on a striped road surface is photographed from above, and an edge of the vehicle is extracted based on the density difference between the vehicle and the road surface in this vehicle image, and the vehicle width is calculated. Measuring.
Japanese Patent Laid-Open No. 9-54892

However, in the above-described technique, the measurement accuracy of the outer dimensions is low. In other words, in the above-described technique, the edge of the vehicle is extracted based on the density difference between the vehicle and the road surface, so that noise for the road surface density such as shadows on the road surface or mud dirt is attached. If this occurs, the vehicle edge cannot be extracted based on the density difference. Therefore, the vehicle width is erroneously detected.
The present invention has been made in view of these points, and provides a technique capable of measuring the external dimensions of a vehicle with high accuracy without being affected by noise on the road surface such as shadows and mud. With the goal.

  In order to achieve the above object, in the present invention, a one-dimensional image acquisition unit captures a vehicle crossing the front of the background to generate a two-dimensional image including the background and the vehicle image, and based on the image not including the vehicle. A vehicle image is extracted by comparing the identified reference image with another image. When the image of the vehicle is extracted, the outer dimensions of the vehicle are measured based on the image.

  That is, in the one-dimensional image acquisition means, a fixed predetermined range is set as a photographing range, and a two-dimensional image is generated by photographing at a predetermined sampling period. Therefore, if the vehicle does not cross the shooting range, the same shooting object is continuously shot. For this reason, the image of the part where the vehicle does not pass is always the same subject of photography, and is always the same image on the two-dimensional image. Therefore, if a reference image is specified based on a part that does not include a vehicle on a two-dimensional image and compared with other images, the difference between the two increases when the vehicle is included in the comparison target. When they are not included, the difference between them is almost “0”. Therefore, when the difference between the two exceeds a predetermined threshold, the image of the vehicle can be extracted by extracting the image.

  In the present invention, since the one-dimensional image capturing range is imaged by the one-dimensional image acquisition means, even if noise such as a shadow is present on the background portion, it is not affected at all by these. An image of the vehicle can be extracted. That is, even if noise such as a shadow is present on the background portion while the vehicle is photographed by the one-dimensional image acquisition means, the same noise is always generated in the photographing result.

  Therefore, even if the reference image including the noise is defined, the image is always the same as the reference image in the portion not including the vehicle, and the vehicle and other images (background or Noise). For this reason, it is possible to extract a vehicle image without being affected by shadows at all, and based on this image, the vehicle's external dimensions (vehicle length (length in the traveling direction of the vehicle body), vehicle height (road surface) The vehicle body length in the direction perpendicular to the vehicle body) and the vehicle width (length in the vehicle body left-right direction)) can be measured.

  The one-dimensional image acquisition unit only needs to be able to photograph a predetermined range in at least a direction perpendicular to the traveling direction of the vehicle. For example, a color line sensor camera (hereinafter referred to as a line camera) can be employed. Of course, a two-dimensional sensor may be used to extract a photographing result in a specific direction from the image, and various configurations can be employed. However, in general, when a line sensor is used, processing for acquiring an image can be performed at a higher speed than when a two-dimensional sensor is used, and an image can be acquired with a short sampling period.

  In the present invention, since the vehicle image is extracted by comparing the reference image with another image, it is preferable to acquire the image so that the difference between the two images easily appears. For example, the color information can be acquired. If configured, it is possible to compare colors (lightness, hue, saturation), and to perform measurement with higher accuracy than in the case of comparison only with shading. Various sensors can be used as sensors for acquiring color information, such as image sensors with three color signal outputs such as RGB (red, green, blue) and CMY (cyan, magenta, yellow). Is possible. Of course, the color is not limited to the above-described color, and for example, a sensor that forms one pixel by adding a green signal output to the three colors of CMY may be used. As the color information, the information such as RGB may be compared, or the lightness, hue, and saturation may be calculated from the information such as RGB to obtain a color difference described later. Alternatively, it is possible to employ a sensor that directly acquires brightness, hue, and saturation by acquiring various optical data such as spectral reflectance for each pixel.

  Furthermore, in the one-dimensional image acquisition means, it is only necessary to capture a predetermined range in a direction perpendicular to the traveling direction of the vehicle. Therefore, an image sensor is arranged at a position that does not interfere with the traveling of the vehicle, and in its field of view. It only needs to be installed so that the background is included. Therefore, it is possible to adopt a configuration in which sensors are installed on a pillar installed on the side of the road surface or a beam extending in the road surface direction from the top of the column installed on the side of the road surface.

  The background portion may be installed in the shooting range of the one-dimensional image acquisition unit and may be the background of the vehicle image when the vehicle travels on the road surface. Therefore, the background may be installed on the road surface, may be installed on a pillar extending in a direction perpendicular to the road surface, or may be installed on both sides. The two-dimensional image generation means only needs to be able to form a two-dimensional image from data accumulated by photographing a vehicle crossing a one-dimensional photographing range, and obtain and obtain a one-dimensional image a plurality of times at a predetermined sampling period. A two-dimensional image may be generated by arranging the one-dimensional images in order.

  The image extracting means only needs to be able to extract the vehicle image by comparing the reference image with another image. The reference image only needs to be specified based on a portion that does not include the vehicle image, and may be either the background or a shadow on the background, or both. As described above, noises such as shadows included in the shooting range remain at the same position during shooting, and thus the same image continues in the horizontal direction on the two-dimensional image. Therefore, the reference image is preferably defined with the same width as the entire one-dimensional imaging range. For example, a specific line that does not include the vehicle image may be used as the reference image, or an average of a plurality of lines that do not include the vehicle image may be used as the reference image, and various configurations may be employed.

  The threshold for extracting the vehicle image only needs to indicate a difference for distinguishing the vehicle from other backgrounds (including noise). For example, a one-dimensional image acquisition unit captures an image of the same object. It is possible to adopt a configuration in which the absolute value T of the magnitude of the shooting error that occurs when the threshold value is used as a threshold and the comparison target is a vehicle when the absolute value of the difference exceeds T.

  Further, the measuring means only needs to be able to calculate any one or a combination of the vehicle length, the vehicle height, and the vehicle width from the image of the vehicle. That is, since the number of pixels in the traveling direction corresponds to the length of the vehicle in the vehicle image, the vehicle length may be calculated from the number of pixels in the traveling direction. Further, the vehicle height may be calculated from the number of pixels in the direction perpendicular to the traveling direction of the vehicle, and the vehicle width may be calculated from the number of pixels in the left-right direction of the vehicle.

  Further, as a preferable configuration when comparing the reference image with another image, the reference image is defined in front and rear of the vehicle, and in the two-dimensional image, scanning from the front side to the rear side of the vehicle and scanning from the rear side to the front side are performed. It is possible to adopt a configuration for performing the above. That is, if shooting is started before the vehicle passes through the shooting range by the one-dimensional image acquisition means, there is a portion that does not include the vehicle in front of the vehicle in the two-dimensional image. 1 reference image is set. When comparing the first reference image with another image, a vehicle image is extracted by scanning the two-dimensional image so as to change the line to be sequentially compared from the front side to the rear side of the vehicle. be able to.

  In addition, if the photographing is finished after a predetermined time has elapsed after the vehicle has passed through the photographing range by the one-dimensional image acquisition means, there is a portion that does not include the vehicle behind the vehicle in the two-dimensional image. A second reference image is set based on the image. When comparing the second reference image with another image, a vehicle image is extracted by scanning the two-dimensional image so as to change the line to be sequentially compared from the rear side to the front side of the vehicle. be able to.

  According to the above configuration, it is possible to accurately extract an image of the vehicle even when the state of noise in the imaging range varies due to the passage of the vehicle. For example, if a vehicle with mud on the rear wheel passes through the shooting range, and mud remains on the road surface constituting the background of the shooting range after passing the rear wheel, the rear wheel is A mud image will be included on the side. However, since the second reference image is defined based on the image on the rear side of the vehicle, the second reference image includes this mud image and is scanned from the rear side to the front side of the vehicle. Then, it is possible to extract a vehicle in which a difference from mud is generated in the image.

  Further, when the state of the mud changes due to the vehicle passing through the road surface on which mud adheres before the vehicle passes, in the two-dimensional image, the state of the road surface image changes before and after the vehicle passes. However, scanning from the front side to the rear side of the vehicle with the first reference image described above, and scanning from the rear side to the front side of the vehicle with the second reference image makes it possible to influence the mud on the road surface. It is possible to accurately extract an image of the vehicle without receiving it.

  The background described above is preferably a specific chromatic color, and when the difference is evaluated by the image extracting means, it is preferable to evaluate the difference between the reference image and other images including the color difference. That is, if the background is composed only of a specific chromatic color, the difference between the reference image and the vehicle becomes clear, and the vehicle image can be easily extracted. Chromatic colors are not limited, but can be selected from any color, so if you use a chromatic color that is less frequently used as a vehicle color as a background, you can perform image processing with very little misrecognition. Become.

  Here, as the chromatic color, it is preferable to adopt a chromatic color that is used as a vehicle color and is relatively small in comparison with other colors. That is, when the vehicle color and the background color are close, it is difficult to distinguish the background image from the vehicle image based on the difference. However, if the background is configured with a chromatic color that is relatively low in frequency compared to other colors, the frequency of occurrence of a situation in which the background image and the vehicle image cannot be distinguished is extremely high. It is possible to realize a vehicle measurement device that is reduced in size and extremely low in error occurrence frequency.

  In addition, as a chromatic color whose frequency used as the color of the vehicle is relatively small compared to other colors, for example, take statistics on the color of the vehicle running in the country where the vehicle measurement device is operated, A color having a frequency of “0” or a frequency of being used may be smaller than that of other colors. If the frequency used is “0”, the error frequency can be substantially set to “0”. In order to distinguish colors, colors that exceed a predetermined color difference may be defined as different colors.

  Here, when evaluating the difference between the reference image and another image, the color difference including all of the brightness, hue, and saturation may be evaluated, or any one of the brightness, hue, and saturation may be evaluated. Alternatively, evaluation may be performed using any two combinations, but it is preferable to evaluate a difference including a hue. That is, in the background color and the vehicle color, the hue is difficult to change depending on the weather and surrounding brightness, and when evaluating the frequency of use as the color of the vehicle, the color difference based on the hue is evaluated. The difference becomes clearer. Therefore, the background and the vehicle can be accurately distinguished by evaluating the difference including the hue among the color components.

  Note that the chromatic color forming the background portion may be at least that color in the photographing range of the one-dimensional image acquisition means, and a paint may be applied to the member, a sheet having a surface of this chromatic color, etc. May be affixed to a pillar or the like, or this chromatic color may be displayed by a light emitting element, and various configurations can be employed.

  Assuming that the vehicle measurement device is operated outdoors, the vehicle image crossing between the one-dimensional image acquisition unit and the background is acquired by the one-dimensional image acquisition unit, and the background is removed after the two-dimensional image is generated. When a configuration is adopted in which the background color is extracted and the background is removed or the background is removed based on the difference between the reference image and another image, the vehicle image including the shadow of the vehicle itself is used. It can happen.

  Therefore, in the image from which the background is removed, it is possible to accurately measure the outer dimensions of the vehicle by removing the shadow existing at least outside the end of the vehicle image from the shadow of the vehicle. That is, when calculating the external dimensions from the vehicle image, the vehicle length is determined from the front end and the end in the vehicle traveling direction, the vehicle height is determined from the front end and the end in the vehicle height direction, and the vehicle left and right are calculated. The vehicle width is calculated from the end of the direction.

  Therefore, if there is a shadow image outside the vehicle image (outside the vehicle front-rear direction, height direction, and left-right end), the position of the tip, end, or end is erroneously affected by the shadow. There is a possibility of recognition. However, if at least the shadow that exists outside the end of the vehicle is removed, there is no possibility that this erroneous recognition will occur, and the external dimensions of the vehicle can be accurately measured.

  As a preferable configuration example when removing the shadow as described above, the lightness is lower than the background color and / or the lower saturation color, and further, the lightness is higher than the tire color or the higher saturation or both. It is possible to adopt a configuration that removes the shadow that is the color of the. That is, since the shadow of the vehicle is formed on the background, it has a darker color than the background (lower brightness than the background, lower saturation than the background, or lower brightness and lower saturation than the background).

  On the other hand, the tire in a vehicle is usually black, and although it is a little, it is low brightness or low saturation or low brightness and low saturation from the peripheral part of the shadow on a background. Therefore, if it is assumed that the color of the lightness and / or saturation is higher than the color of the tire (for example, black), the shadow can be removed without removing the tire. It should be noted that when removing the shadow, it is sufficient to remove at least the shadow existing outside the edge of the vehicle image, so it is sufficient to remove only the peripheral edge of the shadow.

  At the periphery of the shadow, the color is relatively close to the color of the background, and at the center of the shadow is relatively far from the color of the background and close to black. Therefore, as described above, the peripheral portion of the shadow is formed by removing the lightness and / or low saturation or both of the colors of the background portion and the lightness and / or high saturation or both of the colors of the tire. Can be reliably removed, and erroneous recognition can be prevented when measuring the outer dimensions of the vehicle.

  Furthermore, in the configuration in which the background is extracted based on the color of the shadow formed on the background as described above, it is preferable to evaluate the color while eliminating the influence of weather and time as much as possible. Therefore, a part of the background portion is configured as a reference color portion made of a known color, and the color of the entire image is set so that the image of the reference color portion becomes the known color in the two-dimensional image acquired as described above. May be corrected. This correction eliminates color casts (such as redness due to the color of the sunset or a decrease in brightness at a later time) as the reference color part approaches a known color, and the color in the background image is always constant. Can be close to the color.

  As a result, it is possible to remove the background and shadows very easily. In particular, when removing the vehicle shadow based on the threshold when removing the vehicle shadow based on the color darker than the background, the above-mentioned correction is performed to appropriately remove the shadow edge. It becomes possible to remove.

  Furthermore, the measuring means may measure the outer dimensions of the vehicle based on the vehicle image. Further, when calculating the vehicle length, it is possible to measure the vehicle length based on the speed of the vehicle. In this case, the speed of the vehicle is acquired by a speed measuring unit that measures the speed of the vehicle, and the length in the vehicle traveling direction in one line is calculated from the sampling period of the one-dimensional image acquisition means and the speed of the vehicle. The vehicle speed measuring unit only needs to be able to acquire the vehicle speed, and various sensors can be employed.

  In addition, since the vehicle measurement apparatus mentioned above advances a process in the procedure peculiar to this application, it is realizable also as invention of the method characterized by the procedure. Moreover, it is realizable also as invention of the program for making a computer implement | achieve the procedure. Of course, the vehicle measurement device, method, and program may be realized as a part of another device, method, and program, or may be realized by combining a plurality of devices, methods, and programs, Various modes can be adopted. Of course, the present invention can be realized as a recording medium on which the program is recorded.

Hereinafter, embodiments of the present invention will be described in the following order.
(1) Configuration of the vehicle measurement device:
(2) Vehicle measurement processing:
(2-1) Vehicle image extraction processing:
(3) Other embodiments:

(1) Configuration of the vehicle measurement device:
FIG. 1 is a perspective view showing a vehicle measuring apparatus 10 according to an embodiment of the present invention. The vehicle measuring device 10 includes a sensor attached to a pillar installed on the side of a road surface, a background portion serving as a background of the sensor, and a computer that performs data processing. That is, a plurality of pillars 20a to 20e are installed on both sides of the road surface 11 on which the vehicle to be measured can travel, and sensors are attached to these pillars 20a to 20e. Further, a beam 20f is attached to the column 20c, and a sensor is also attached to the beam 20f. Further, a pillar 21 is erected on the opposite side of the pillar 20c across the road surface 11, and the road surface 11 between the one side of the pillar 21 on the road surface side and the pillars 21 and 20c is colored chromatic (for example, green). Cage background portions 21a and 21b are formed.

  Approach sensors 30a and 30b, which are a pair of sensors, are attached to the pillars 20a and 20b, respectively. The ingress sensor 30b outputs electromagnetic waves (visible light, infrared rays, etc.) toward the ingress sensor 30a, and the ingress sensor 30a detects the electromagnetic waves. Therefore, when the vehicle 12 travels on the road surface 11 and blocks the electromagnetic wave from the entry sensor 30b, the electromagnetic wave is not detected by the entry sensor 30a, and it can be detected that the vehicle 12 has entered the measurement system of the vehicle measurement device 10. .

  The height ha of the approach sensor 30a from the road surface 11 is different from the height hb of the approach sensor 30b from the road surface 11. Therefore, the electromagnetic wave from the approach sensor 30b crosses the road surface 11 while traveling from the upper side to the lower side. As a result, when the vehicle 12 travels on the road surface 11 at any height, the electromagnetic wave from the approach sensor 30b is shielded, and compared with the case where the approach sensor arranged at the same height is used. In addition, it is possible to detect the approach of the vehicle 12 having a wide range of heights.

  Retraction sensors 30c and 30d made up of a pair of sensors are attached to the columns 20d and 20e, respectively. The exit sensors 30c and 30d have the same configuration as the entrance sensors 30a and 30b. The exit sensor 30d outputs an electromagnetic wave toward the exit sensor 30c, and the exit sensor 30c detects the electromagnetic wave. Accordingly, when the vehicle 12 travels on the road surface 11 and the electromagnetic wave from the exit sensor 30d once blocked reaches the exit sensor 30c, it can be detected that the vehicle 12 has passed through the measurement system of the vehicle measurement device 10. The exit from the vehicle measurement device 10 can be detected.

  Also in the exit sensors 30c and 30d, the height hc of the exit sensor 30c from the road surface 11 is different from the height hd of the exit sensor 30d from the road surface 11. Therefore, it is possible to detect the exit of the vehicle 12 having a wide range of heights compared to the case of using the exit sensor arranged at the same height. The approach sensors 30a and 30b and the exit sensors 30c and 30d are also used to detect the speed of the vehicle 12, and the distance D between the sensors 30a and 30c is known and the height ha of the approach sensor 30a is determined. The exit sensor 30c is installed such that the height hc is the same height, and the height hb of the entry sensor 30b and the height hd of the exit sensor 30d are the same height.

  The column 20c extends perpendicularly to the road surface 11, and a beam 20f extending parallel to the road surface and extending toward the road surface is attached to the upper end of the column 20c. Line sensors 31a to 31d for photographing the vehicle 12 are attached to the pillars 20c and the beams 20f. That is, two line sensors 31a and 31b are attached to the pillar 20c, and two line sensors 31c and 31d are attached to the beam 20f, and a vehicle traveling on the road surface 11 is photographed by these line sensors 31a to 31d.

  The line sensors 31a to 31d in the present embodiment are sensors having a long field of view in one direction, and three color component values (in this embodiment, RGB gradations) for each of a plurality of pixels arranged in one direction. Value). In the present embodiment, one line of data is acquired at every predetermined time interval (sampling period). The visual fields of the line sensors 31a and 31b attached to the column 20c exist on a common straight line, and this straight line coincides with the central straight line 21a1 of the background portion 21a and the central straight line 21b1 of the background portion 21b. Therefore, the line sensors 31a and 31b can take images of the predetermined range on the plane including the straight lines 21a1 and 21b1 as a visual field.

  The field of view of the line sensors 31c and 31d attached to the beam 20f also exists on a common straight line, and this straight line coincides with the central straight line 21b1 of the background portion 21b and the central straight line 21a1 of the background portion 21a. Accordingly, the line sensors 31c and 31d can take an image of the predetermined range on the plane including the straight lines 21a1 and 21b1 as a field of view.

  In addition, the plane including the straight lines 21a1 and 21b1 is perpendicular to the traveling direction G of the vehicle 12, and the line sensors 31a and 31b are directed to the pillar 21, so that the line sensors 31a and 31b move along the road surface 11. The side surface of the passing vehicle 12 is photographed. Since the line sensors 31 c and 31 d are directed toward the road surface 11, the line sensors 31 c and 31 d take an image of the upper surface of the vehicle 12 passing through the road surface 11.

  Furthermore, since both line sensors 31a and 31b photograph the side of the vehicle and photograph the same object, the photographed images of the respective line sensors 31a and 31b are combined after correcting the aberration of the lens. Similarly, since the line sensors 31c and 31d both photograph the upper surface of the vehicle, the photographed images of the line sensors 31c and 31d are combined after correcting the lens aberration.

  It is preferable that the approximate center of the visual field of the line sensor 31a is configured to substantially coincide with the lower end of the vehicle 12, and the approximate center of the visual field of the line sensor 31b is substantially aligned with the upper end of the vehicle 12. It is preferable that the left end of the vehicle 12 and the approximate center of the visual field of the line sensor 31 d substantially coincide with the right end of the vehicle 12. With this configuration, the vehicle 12 can be configured such that errors due to lens aberration are reduced at the top, bottom, left, and right ends of the vehicle 12.

  In this embodiment, the vehicle measuring device 10 includes a computer 40, and the computer 40 is connected to the line sensors 31a to 31d, the entrance sensors 30a and 30b, and the exit sensors 30c and 30d via cables. . Accordingly, the computer 40 can acquire the output signals of these sensors and perform analysis based on the output signals.

  FIG. 2 is a block diagram illustrating a configuration of the computer 40. As shown in FIG. 2, the computer 40 includes a CPU 41, a ROM 42, a RAM 43, an HDD 44, and I / Fs 45a to 45c. An I / F 45a is an interface between the line sensors 31a to 31d, the ingress sensors 30a and 30b, and the egress sensors 30c and 30d. The CPU 41 controls each sensor via the I / F 45a and acquires an output signal thereof.

  The I / F 45b is a connection interface with the display 40a, and the CPU 41 outputs data for performing various displays to the display 40a via the I / F 45b. The display 40a acquires this data and performs various displays. The I / F 45c is an interface with an input device, and the CPU 41 acquires signals from the keyboard 40b and the mouse 40c and grasps the operation contents on the keyboard 40b and the mouse 40c.

  The CPU 41 can execute programs recorded in the ROM 42 and the HDD 44 using the RAM 43 as a work area. In the present embodiment, a vehicle measurement program can be executed as one of the programs, and the vehicle measurement program includes modules of a sensor control unit 41a, an image extraction unit 41b, and a vehicle measurement unit 41c. This vehicle measurement program is a program for measuring the vehicle length, vehicle height, and vehicle width of the vehicle 12 on the basis of various data recorded in advance in the HDD 44. Image data is generated during the execution process and analyzed. Do.

  That is, the inter-sensor distance data 44e, sampling cycle data 44f, height line length data 44g, and width line length data 44h are recorded in the HDD 44 in advance. The HDD 44 also records data generated in the course of executing the vehicle measurement program. In this embodiment, the one-dimensional image data 44a, the two-dimensional image data 44b, the reference image data 44c, and the vehicle image data 44d are executed. Generated and recorded in the process.

  The inter-sensor distance data 44e is data indicating the distance D between the sensors 30a and 30c. The sampling cycle data 44f is data indicating the sampling cycle of the line sensors 31a to 31d. The height line length data 44g is data indicating a length corresponding to one pixel in the height direction in a one-dimensional image photographed by the line sensors 31a and 31b, and a reference sample having a known length is photographed. Is acquired in advance. The width line length data 44h is data indicating a length corresponding to one pixel in the width direction (the left-right direction of the vehicle body) in a one-dimensional image captured by the line sensors 31c and 31d. It is acquired in advance by photographing a reference sample.

  The sensor control unit 41a is a module that controls the line sensors 31a to 31d, the entrance sensors 30a and 30b, and the exit sensors 30c and 30d. Acquire an image. The image extraction unit 41b includes a reference image setting unit 41b1 and a color difference determination unit 41b2, and performs a process of generating a two-dimensional image based on the captured image and extracting a vehicle image from the two-dimensional image. The vehicle measurement unit 41c includes a vehicle length calculation unit 41c1, a vehicle height calculation unit 41c2, and a vehicle width calculation unit 41c3. The vehicle length calculation unit 41c1 uses the vehicle length and vehicle height calculation unit 41c2 to execute the vehicle. The vehicle height and vehicle width calculation unit 41c3 calculates the vehicle width of the vehicle 12 and displays it on the display 40a.

(2) Vehicle measurement processing:
Next, processing performed by each unit of the vehicle measurement program in the above configuration will be described in detail. FIG. 3 is a flowchart showing the processing of the vehicle measurement program. As shown in the figure, when the vehicle measurement program is executed, the sensor control unit 41a determines whether or not the ingress sensor 30a is on based on a signal from the ingress sensor 30a (step S100). Here, the state in which the vehicle 12 blocks the electromagnetic wave from the ingress sensor 30b and the electromagnetic wave does not reach the ingress sensor 30a is referred to as the ingress sensor 30a being on.

  The sensor control unit 41a continues to detect signals from the ingress sensor 30a until it is determined in step S100 that the ingress sensor 30a has been turned on. When it is determined that the ingress sensor 30a has been turned on, the line sensors 31a to 31a. Photographing by 31d is performed (step S105). Here, the sensor control unit 41a acquires data of a one-dimensional image captured by each of the line sensors 31a to 31d during a predetermined sampling period, and records it in the HDD 44 as one-dimensional image data 44a.

  Next, the sensor control unit 41a determines whether or not the exit sensor 30c is on based on the signal from the exit sensor 30c (step S110). Here again, the state in which the vehicle 12 blocks the electromagnetic wave from the exit sensor 30d and the electromagnetic wave does not reach the exit sensor 30c is referred to as the exit sensor 30c being on.

  When it is determined in step S110 that the exit sensor 30c is on, since the vehicle 12 has reached between the exit sensors 30c and 30d, the sensor control unit 41a determines that the entrance sensor 30a is on in step S100. The time T from when the exit sensor 30c is turned on to when the exit sensor 30c is turned on is acquired (step S115). That is, since the distance D between the approach sensor 30a and the exit sensor 30c is known, the time T is acquired in order to calculate the speed of the vehicle 12 based on the distance D. In addition, since this process should be performed once per vehicle, if the time T has been acquired for each vehicle, step S115 is skipped. Further, a counter C (default value is 0) for determining the exit of the vehicle 12 is set to 1 (step S117).

  When it is not determined in step S110 that the exit sensor 30c is on, and after executing step S117, the sensor control unit 41a determines whether the exit sensor 30c is off and the counter C is 1 (step). S120). Here, when the vehicle 12 finishes passing between the exit sensors 30c and 30d, and the electromagnetic wave from the exit sensor 30d blocked by the vehicle 12 reaches the exit sensor 30c again, the exit sensor 30c is off. Call. That is, since the counter C is 1 when the vehicle 12 reaches the exit sensor 30c, the completion of the passage of the vehicle 12 is determined by determining that the counter C is 1 and the exit sensor 30c is off. ing.

  If it is not determined in step S120 that the exit sensor 30c is off and the counter C has become 1, the process returns to step S105, and the line sensors 31a to 31d continue to acquire the one-dimensional image data 44a. Therefore, the sensor control unit 41a captures and accumulates the one-dimensional image data 44a from the time when the vehicle 12 enters between the entry sensors 30a and 30b until the vehicle 12 has passed between the exit sensors 30c and 30d. At the same time, the process of acquiring the time T is performed.

  If it is determined in step S120 that the exit sensor 30c is turned off, the image extraction unit 41b generates two-dimensional image data 44b from the one-dimensional image data 44a (step S125). That is, since the one-dimensional image data 44a is a one-dimensional image photographed by the line sensors 31a to 31d, first, the one-dimensional images photographed at the same sampling timing are synthesized. That is, since the captured images by the line sensors 31a and 31b are images of the side surface of the vehicle 12, the one-dimensional images are combined so that the pixels in the common part are overlapped with each other. Similarly, since the captured images by the line sensors 31c and 31d are images on the upper surface of the vehicle 12, the one-dimensional images are combined so that the pixels in the common part are overlapped with each other. Various known techniques can be employed for this connection.

  When the one-dimensional image of the side surface and the one-dimensional image of the upper surface of the vehicle 12 are generated as described above, the image extraction unit 41b further combines these one-dimensional images to generate a two-dimensional image. That is, since the one-dimensional image of the vehicle side surface and the one-dimensional image of the upper surface of the vehicle having different sampling timings are obtained after the combination is performed, these are arranged in the order of sampling. As a result, a two-dimensional image of the side surface and a two-dimensional image of the upper surface of the vehicle 12 are generated.

  FIG. 4A is a diagram showing a two-dimensional image of the side surface of the vehicle 12 generated in step S125. Since the line sensors 31a and 31b in the present embodiment continue to capture the background portions 21a and 21b, the images acquired by the line sensors 31a and 31b when the vehicle is not passing are the straight lines 21a1 and 21b1 of the background portions 21a and 21b. . On the other hand, when the vehicle 12 is passing, images of the vehicle 12 with the background of the straight lines 21a1 and 21b1 of the background portions 21a and 21b are acquired by the line sensors 31a and 31b.

  However, when a shadow is generated in the background, for example, when a shadow is generated in a part of the background portion 21a, the shadow does not move in a short time when the vehicle 12 passes through the imaging range of the line sensors 31a and 31b. Therefore, the shadow image 21a2 is formed at the same height in the two-dimensional image. In this case, the shadow image 21a2 may remain together with the image of the vehicle 12 only by removing the same color portion as the background from the two-dimensional image. Therefore, in the present embodiment, processing for extracting an image of the vehicle without being affected by the shadow is performed (step S130).

(2-1) Vehicle image extraction processing:
FIG. 5 is a flowchart showing the vehicle image extraction process in step S130. In the present embodiment, a two-dimensional coordinate (i, j) is defined for each pixel in the two-dimensional image, and the numerical value of the coordinate j increases from top to bottom in the two-dimensional image. On the other hand, the increase direction of the coordinate i changes depending on the order (scan direction) of extracting the image to be compared with the reference image.

  That is, in the present embodiment, scanning is performed in both the scan direction for scanning from the front side to the rear side of the vehicle in the two-dimensional image and the scan direction for scanning from the rear side to the front side of the vehicle, The definition of coordinate i and the increasing direction are reversed in each scan direction. It should be noted that the front side of the vehicle is an image that is sampled earlier in the two-dimensional image, and the sampled timing is delayed as it becomes the rear side. The value range of the coordinate i is 0 to imax, and the value range of the coordinate j is 0 to jmax.

  In this coordinate system, first, the reference image setting unit 41b1 determines a scanning direction, and sets two-dimensional coordinates (i, j) for specifying an image to be compared to an initial value (1, 0). (Step S200). Further, one line where the coordinate i is “0” is set as the reference image (step S205). That is, in this embodiment, the scan direction is set before starting the scan, and different reference images are set for each.

  Specifically, in the two-dimensional image shown in FIG. 4A, a scan direction from the front side to the rear side and a scan direction from the rear side to the front side are executed. Decide either. At this time, the increasing direction of the coordinate i is also determined. For example, when the scan is from the front side to the rear side, the coordinate i is defined to increase from the front side to the rear side, and when the scan is from the rear side to the front side, the coordinate i is from the rear side to the front side. Define to increase toward.

  As described above, the reference image is an image for one line, and the coordinate i is “0”. Therefore, in this embodiment, in the two-dimensional image generated in step S125, the image for one line acquired first or the image for one line acquired last is the reference image. 4A shows a reference image S for scanning from the front side to the rear side. At this time, the initial value (1, 0) of the two-dimensional coordinates (i, j) is shown. Indicates the pixel present at the top in the line on the right side of the reference image S. Further, if the coordinate i is increased by scanning, the position of the line is sequentially shifted to the right side.

  In the present embodiment, the image data of the reference image S set as described above is recorded in the HDD 44 as the reference image data 44c. Needless to say, in the reference image S, any line that does not include a vehicle may be defined as the reference image S from the two-dimensional image as long as an image serving as a reference for comparison may be defined. A reference image may be defined by extracting a plurality of lines that are not present and taking an average, or a reference image may be generated in advance before generating a two-dimensional image, and various configurations may be employed. .

  When the scan direction, the initial coordinate value, and the reference image are set as described above, a scan for sequentially comparing the reference image with another image is performed. For this purpose, first, the color difference determination unit 41b2 extracts one pixel (referred to as a reference pixel) from the reference image, and calculates a color difference ΔE from a pixel to be scanned (referred to as a scan target pixel) ( Step S210). Here, the reference pixel is one of the pixels included in one line of the reference image S, and is specified by coordinates (0, j). On the other hand, a pixel to be scanned is a pixel included in a line to be scanned, and is specified by coordinates (i, j). That is, the reference pixel and the scan target pixel have different lines, but the vertical position (the value of coordinate j) matches.

When the color difference ΔE is calculated, it is determined whether or not the color difference ΔE is larger than a predetermined threshold T ₁ (step S215). Here, the threshold value T ₁ is a threshold value for determining whether or not the reference pixel and the scan target pixel are substantially the same image, and is substantially “0”. It is a value that gives a margin for shooting error. Therefore, by determining whether or not the color difference ΔE is greater than the threshold value T _1, it is possible to determine whether or not the reference pixel and the scan target pixel are substantially the same, and the scan target pixel is the background or the vehicle. Can be determined.

  Specifically, as shown in FIG. 4A, the two-dimensional image in the present embodiment is an image of the background portions 21a and 21b, and only when the vehicle 12 crosses the front of the background. An image is obtained. Further, since the reference image S is an image of a portion in which the vehicle 12 is not included in the two-dimensional image, the reference image S is composed only of the background. Therefore, the reference image S is compared with other images, and if the image is different from the reference image S, the image can be specified as the image of the vehicle 12.

For example, in the image for one line at the coordinate i ₁ shown in FIG. 4A, the image of the vehicle 12 is not included, so even if the color difference ΔE with respect to the reference image S is calculated, the value is almost “0”. a is, it does not exceed the threshold T _1. On the other hand, in the one line of the image at the coordinates i _2, it includes an image of the vehicle 12, the color difference ΔE between the image and the reference image S of the vehicle 12 exceeds a threshold value T _1. Therefore, by calculating the color difference ΔE, it is possible to determine whether the image is the vehicle 12 or not.

Therefore, at step S215, the when the color difference ΔE is determined to a predetermined larger thresholds T ₁ is the coordinates (i, j) is set to the effect that the vehicle (step S220). Further, at step S215, the when the color difference ΔE is not determined to be greater than thresholds T ₁ which is determined in advance, the coordinates (i, j) is set to the effect that the background (step S225). These settings may be made, for example, by associating a preset flag with the coordinates (i, j).

  When setting that the vehicle or the background is the coordinate (i, j), the color difference determination unit 41b2 increments the value of the coordinate j (step S230), and whether or not the coordinate j is larger than the maximum value jmax. Is discriminated (step S235). If it is not determined in step S230 that the coordinate j is greater than the maximum value jmax, the processes in and after step S210 are repeated.

  When it is determined in step S235 that the coordinate j is larger than the maximum value jmax, the comparison for one line is completed, so the color difference determination unit 41b2 further increments the coordinate i and sets the coordinate j to “0”. (Step S240), and it is determined whether or not the coordinate i is larger than ½ of the maximum value imax (step S245). If it is not determined in step S245 that the coordinate i is greater than ½ of the maximum value imax, the processing from step S210 is repeated, and if it is determined that the coordinate i is greater than ½ of the maximum value imax. It is determined whether or not the bi-directional scan is completed (step S250).

  If it is not determined in step S250 that the scanning in the two directions has been completed, the process returns to step S200, the scanning is set in the unprocessed direction, and the processes in and after step S200 are repeated. If it is determined in step S250 that the scanning in the two directions has been completed, the pixel set to be an image of the vehicle 12 is extracted by the above processing, and is set as vehicle image data 44d and recorded in the HDD 44 (step 44). S255). In this embodiment, since scanning is performed from two directions, in step S245, in order to perform necessary and sufficient scanning, the coordinate i has its maximum value imax so that the scanning in each direction ends at the center of the left and right of the image. It is discriminated whether or not it is greater than 1/2.

  In FIG.4 (b), the image of the vehicle 12 extracted by the above-mentioned process is shown, and the background removed when extracting the image of the vehicle 12 is shown by hatching. The vehicle image data 44d after removing the background may include only the image of the vehicle 12, but in consideration of the ease of handling of data by the computer 40, dummy data ( For example, a configuration in which a predetermined color such as white or background color is arranged may be employed.

  According to the above processing, not only when the background portions 21a and 21b have a uniform color, but also when shadows are formed on the background portions 21a and 21b or dirt such as mud adheres, The image of the vehicle 12 can be extracted. For example, when a shadow is formed on the background portion 21a, the shadow hardly changes in a short time when the vehicle 12 passes through the imaging range of the line sensors 31a and 31b. As shown, a shadow image 21a2 that is uniform in the horizontal direction is formed.

Accordingly, the shadow image 21a2 is also included in the reference image S, and in comparison between the reference image S and another image, the shadow image 21a2 on one line whose coordinate i is “0” and the other image. Will be compared. In this comparison, when the pixel to be scanned is not an image of the vehicle 12, since both the reference image S and the other image are shadow images, the color difference ΔE is almost “0”, and the pixel to be scanned is if the image of the vehicle 12, the color difference ΔE exceeds the threshold value T _1. As a result, it is possible to appropriately extract the image of the vehicle 12.

  Furthermore, in this embodiment, even when the state of mud dirt or the like in the background portions 21a and 21b changes as the vehicle 12 passes through the shooting range, the image of the vehicle 12 can be appropriately extracted. FIG. 6 is an explanatory diagram for explaining this situation. FIG. 6A shows a case where mud adhering to the rear wheel of the vehicle 12 remains on the background portion 21b (road surface) after the vehicle 12 passes. 2 shows an example of a two-dimensional image.

As shown in the figure, when mud adhering to the rear wheel of the vehicle 12 remains on the road surface, on the two-dimensional image, uniform mud dirt 21b2 is generated in the lateral direction on the rear side of the rear wheel. become. However, in this image, when scanning from the rear side to the front side of the vehicle 12 is performed, the right end image is sequentially used as the reference image S. At this time, the color difference ΔE is substantially “0” from the rear side to the front side of the vehicle 12 in the portion where the mud stain 21b2 exists, but the color difference ΔE is a threshold value when the pixel to be scanned becomes a tire image. beyond T _1, it is extracted as the image of the vehicle 12.

  Note that when a reference image S including mud dirt 21b2 is set and scanning is performed from the rear side to the front side of the vehicle 12, the background on the front side of the rear wheels may be extracted as an image of the vehicle 12. However, when measuring the outer dimensions of the vehicle in the present embodiment, as will be described later, since the end of the image of the vehicle 12 is used, the pixels between the wheels are extracted as the image of the vehicle 12. However, there is no influence on the measurement of the outer dimensions of the vehicle. On the other hand, on the front side of the vehicle, scanning is performed based on the reference image S set on the front side of the vehicle. In this reference image S, the background portion 21b does not include mud dirt 21b2, so the vehicle 12 and the background portion 21b are An image of the vehicle 12 can be extracted with a clear distinction.

  The same applies to the case where mud dirt or the like adhering to the background portion 21b on the road surface changes due to the passage of the vehicle 12, and an image of the vehicle 12 can be appropriately extracted. In this case, as shown in FIG. 6B, a uniform mud dirt 21b3 is generated in the lateral direction on the front side of the vehicle, and the mud dirt is removed or thinned on the rear side of the vehicle. Has occurred. Also in this image scanning, when scanning from the front side to the rear side, the reference image S including the mud dirt 21b3 is set, so that the tire of the vehicle 12 is appropriately extracted by the scanning. On the other hand, since the reference image S in a state where mud dirt does not exist is set during the scan from the rear side to the front side, the image of the vehicle 12 is appropriately extracted even in the scan in this direction. As described above, in the present embodiment, it is possible to appropriately extract the image of the vehicle 12 even when the background portions 21a and 21b change due to the passage of the vehicle 12.

  After the vehicle image data 44d is created by the above processing, the vehicle measurement unit 41c calculates the vehicle length, vehicle height, and vehicle width based on the vehicle image data 44d. In the present embodiment, first, a vehicle length calculation process is performed (step S135). FIG. 7 is a flowchart showing the calculation process of the vehicle length. In the present embodiment, the vehicle length is measured based on the measured value of the speed of the vehicle 12, and the vehicle length calculation unit 41c1 of the vehicle measurement unit 41c first calculates the vehicle speed (step S300). Here, the vehicle speed is calculated by D / T using the distance D and time T.

  Next, the vehicle length calculation unit 41c1 calculates the length in the vehicle length direction for one line in the line sensors 31a and 31b (step S310). That is, since an image for one line in the line sensors 31a and 31b is taken at the sampling period, one line in the vehicle length direction in the two-dimensional image of the vehicle 12 is between the sampling periods of the line sensors 31a and 31b. This corresponds to the distance that the vehicle 12 moves. Therefore, by multiplying the sampling period and the vehicle speed, the length in the vehicle length direction for one line can be calculated.

  After calculating the length in the vehicle length direction for one line, the vehicle length calculation unit 41c1 detects the front end and the end of the vehicle from the image of the side surface of the vehicle 12 (step S320). That is, in the image of the vehicle 12 shown in FIG. 4B, since the distance L between the front end and the end of the image corresponds to the vehicle length, the front end and the end are detected. For example, in the two-dimensional image of the side surface of the vehicle 12, the leftmost pixel and the rightmost pixel may be extracted. As described above, when the background is set to dummy data, the leftmost pixel and the rightmost pixel of pixels that are not dummy data are extracted.

  When the front end and the end of the vehicle 12 are detected in the two-dimensional image, the number of lines (number of samples) in the vehicle body traveling direction in the vehicle 12 is acquired (step S330), and the vehicle length is calculated based on the number of lines (step S330). Step S340). That is, in step S330, the number of lines from the front end to the end including the line at the front end and the end is acquired. Since the length of one line in the vehicle traveling direction has been calculated in step S310, the vehicle length L can be calculated by multiplying the length of the one line by the number of lines from the leading end to the terminal end.

  When the vehicle length L is calculated as described above, the vehicle height calculation unit 41c2 calculates the vehicle height based on the side image of the vehicle 12 extracted in step S130 (step S140). That is, in the one-dimensional image photographed by the line sensors 31a and 31b, the length corresponding to one pixel in the height direction is described by the height line length data 44g, so the vehicle height calculation unit 41c2 Extracts the top and bottom pixels from the side image of the vehicle 12 and multiplies the number of pixels between them by a length corresponding to one pixel in the height direction (FIG. 4B). ) Of H) is calculated.

  Similarly, the vehicle width calculation unit 41c3 calculates the vehicle width based on the top image of the vehicle 12 extracted in step S130 (step S145). That is, in the one-dimensional images taken by the line sensors 31c and 31d, the length corresponding to one pixel in the width direction is described in the width line length data 44h, so the vehicle width calculation unit 41c3 The leftmost pixel and the rightmost pixel are extracted from the top image of the vehicle 12, and the vehicle width is calculated by multiplying the number of pixels between them and the length corresponding to one pixel in the width direction.

  Since the vehicle length, vehicle height, and vehicle width are calculated through the above processing, in this embodiment, the vehicle measurement unit 41c outputs a control signal via the I / F 45b, and the vehicle length is displayed on the display 40a. , The vehicle height and the vehicle width are displayed (step S150). As a result, the user who uses the vehicle measurement device 10 does not have to perform any artificial work, and can measure the vehicle length, vehicle height, and vehicle width by only traveling on the road surface 11 with the vehicle 12. it can.

(3) Other embodiments:
The above-described embodiment is an embodiment of the present invention, and the embodiment of the present invention is not limited to the above-described embodiment. For example, in order to detect a vehicle image more accurately, processing for removing the shadow of the vehicle, color correction, or the like may be performed. FIG. 8 is a block diagram showing a configuration of an embodiment that performs a process for removing a shadow of a vehicle and a color correction process, and FIGS. 9 and 10 are flowcharts thereof.

  Also in the embodiment shown in FIG. 8, the schematic configuration is the same as that in FIG. 2, and can be realized by almost the same hardware configuration as in FIGS. 1 and 2. However, the background portion 21a includes a reference color portion (not shown) at a position included in the shooting range of the line sensor 31a or 31b. The reference color portion is white when the surroundings of the background portions 21a and 21b have a predetermined brightness, and data when captured by the line sensor 31a or 31b is determined in advance.

  Further, as shown in FIG. 8, the HDD 44 records background color data 440c instead of the reference image data 44c. The background color data 440c is data indicating colors obtained when the background portions 21a and 21b are photographed by the line sensors 31a and 31b when the surroundings of the background portions 21a and 21b have a predetermined brightness. Furthermore, the vehicle measurement program in FIG. 8 includes a background removal unit 410b and a shadow removal unit 411b instead of the image extraction unit 41b in FIG. The background removing unit 410b is a module that performs a process of removing a background from the two-dimensional image, and removes the background based on the background color data 440c. The shadow removing unit 411b is a module that performs a process of removing the shadow of the vehicle, which is an extra image when calculating the outer dimensions of the vehicle from the image from which the background has been removed by the background removing unit 410b.

  The process in the vehicle measurement program is the procedure shown in FIG. 9, and most of the process is the same as the process shown in FIG. 3, except that steps S131, 132, and 134 shown in FIG. 9 are performed instead of step S130 in FIG. Is different. FIG. 11A shows an example of a two-dimensional image obtained in step S125 shown in FIG. 9 when the shadow of the vehicle 12 is generated on the road surface. As shown in the figure, the shadow image 21b4 of the vehicle 12 changes with the progress of the vehicle 12 when the vehicle 12 passes through the imaging range of the line sensors 31a and 31b, so that it is uniform in the horizontal direction on the two-dimensional image. Instead, a shadow image 21b4 corresponding to the shape of the vehicle 12 is formed.

  In addition, in the two-dimensional image shown in FIG. 11A, an image of the reference color portion (21a3 in FIG. 11A) formed on the background portion 21a is shown at the top, and in this example, In addition, the entire color cast (redness due to the sunset or darkening due to the later time) is indicated by hatching of the entire two-dimensional image.

  In the present embodiment, after performing steps S100 to S125 by the sensor control unit 41a, the background removal unit 410b performs a process of removing the background from the two-dimensional image (step S131). That is, since the colors of the background portions 21a and 21b to be removed are defined by the background color data 440c, these colors are extracted and removed from the two-dimensional image. At this time, the image of the reference color portion 21a3 is also removed and the image data is stored.

  According to the above processing, an image of the vehicle 12 should be extracted, but when the shadow of the vehicle 12 is formed on the road surface, such as when the vehicle measurement device 10 is installed outdoors, FIG. The image of the vehicle 12 and the shadow image 21b4 shown in a) remain. As described above, the outer dimensions of the vehicle, for example, the vehicle length, is calculated based on the leading edge and the trailing edge of the image after background removal, so that a shadow image 21b4 of the vehicle 12 is displayed as in the example shown in FIG. If it exists in front of the tip, the tip of the shadow image 21b4 may be extracted.

  Therefore, in the present embodiment, the shadow removing unit 411b performs a process of removing at least the peripheral portion of the shadow image 21b4 existing outside the end of the image of the vehicle 12. For this purpose, first, the color of the image is corrected (step S132). That is, since the color of the image obtained after the process of step S131 may have a color cast as shown in FIG. 11A, the color is corrected based on the image data of the reference color portion.

Specifically, as described above, the data when the reference color portion is photographed by the line sensor 31a or 31b is determined in advance. Therefore, when the image of the reference color portion removed in step S131 is different from this data, Correction is performed based on the ratio. For example, the gradation value of each RGB color in the data predetermined as the reference color portion color is R ₀ , G ₀ , B ₀ , and the gradation value of each RGB color in the reference color portion data removed from the two-dimensional image is R _{When it is 1} , G ₁ and B ₁ , the gradation values of the respective RGB colors in the image after processing in step S131 are multiplied by R ₀ / R ₁ , G ₀ / G ₁ and B ₀ / B ₁ , respectively. Of course, in this correction, the corrected value does not exceed the maximum value and the minimum value of the gradation value range. For this purpose, non-linear correction such as gamma correction may be performed.

  As a result, in the corrected image, correction is performed so that the reference color portion is converted to a predetermined color, and the color cast is eliminated. Therefore, in the present embodiment, a process of removing the shadow image 21b4 from the corrected image is performed. In the present embodiment, the peripheral portion of the shadow image 21b4 is removed using the property of the shadow image 21b4. That is, as shown in FIG. 11B, the shadow image 21b4 is generally dark black at the center, but becomes thinner due to the influence of light diffraction as it approaches the periphery from the center (the background image). Have a color).

Therefore, the shadow removing unit 411b performs the processing shown in FIG. 10 to remove the peripheral portion of the shadow image 21b4, and first obtains background color data 440c indicating the color of the background portion 21b (step S400). Next, the shadow removing unit 411b sets threshold values T ₂ and T ₃ for specifying the peripheral portion of the shadow image 21b4 (step S405). Here, the threshold values T ₂ and T ₃ are values set to remove the peripheral portion of the shadow image 21b4 without removing the tire image in the image of the vehicle 12, and are the lower limit values of the colors to be removed. Is the threshold T ₂ and the upper limit is the threshold T ₃ . That is, the peripheral portion of the shadow image 21b4 to be removed is darker than the color of the background portion 21b acquired in step S400, and is lighter than the tire color.

Therefore, in the present embodiment, a color darker than the color of the background portion 21b is defined as a color having lower brightness and lower saturation than the color of the background portion 21b, and lower brightness and lower saturation than the color indicated by the background color data 440c. Data indicating the color of the degree is set as a threshold T ₃ . Further, the color lighter than the color of the tire is defined as the color of high brightness and high chroma from the color of the tire, the data indicating the color of a high lightness and high chroma from the color of the tire to a predetermined a threshold T _2. Of course, the threshold values T ₂ and T ₃ are gradation values of RGB colors. The predetermined tire color can be determined by various methods such as adopting a tire color statistically acquired from a number of vehicle images.

Further, when determining each of the threshold values T ₂ and T ₃ , it suffices if it can be determined so as to remove a color to be removed and hold a color that should not be removed. For example, the background color data 440c indicates It is possible to adopt various methods such as determining a predetermined margin based on the color and the color of the tire and statistically determining a preferable value as the margin.

When the threshold values T ₂ and T ₃ are determined as described above, the shadow removing unit 411b determines a search area where processing for removing the shadow is to be performed (step S410). Here, it is only necessary to determine a region including the shadow image 21b4, and the search region can be determined by various methods such as setting the lower portion of the vehicle 12 to this region. Of course, if the processing load does not become excessive, all the images obtained in step S131 may be used as the search area.

When the search area is determined, it is determined whether or not the pixel data (color to be compared) is included in the threshold values T _{2 to} T ₃ with any one of the pixels included in the search area as a comparison target (step S415). . If it is determined in step S415 that the color to be compared is included in the threshold values T _{2 to} T ₃ , the color to be compared is removed (step S420), and the color to be compared is set to the threshold value T in step S415. if it is determined to be included in the ₂ through T ₃ skips the step S420.

  After this processing, it is determined whether or not the above comparison has been performed for all the pixels in the search region (step S425). If it is not determined in step S425 that the above comparison has been performed for all the pixels in the search region. The pixel to be compared is changed (step S430), and the processing after step S415 is repeated.

  FIG. 11C is a diagram illustrating an image after the peripheral portion of the shadow image 21b4 is removed by the above processing. As shown in the figure, since the peripheral portion of the shadow image 21b4 is removed, the front end and the end in the traveling direction of the vehicle 12 are images of the vehicle 12, and the upper end and the lower end in the vertical direction of the vehicle 12 are the vehicle. There are 12 images. Accordingly, it is possible to accurately calculate the vehicle length L, the vehicle height H, and the like of the vehicle 12 by the processing after step S135.

In the above processing, the peripheral portion of the shadow image 21b4 is removed by relatively comparing the background portion 21b and the tire color with the color of the shadow image 21b4. By performing this correction, shadows can be removed more accurately. That is, by performing color correction, the threshold values T ₂ and T ₃ for removing the peripheral portion of the shadow image 21b4 without removing the tire image in the image of the vehicle 12 can be easily set (for example, in statistical knowledge). Can be set).

  Of course, the embodiment of the present invention is not limited to the above-described example. For example, the color correction process in step S132 in FIG. 9 may be performed before the background removal process in step S131. You may perform the same process as step S130 shown in FIG. In step S400, it is only necessary to acquire data indicating the color of the background portion 21b. Therefore, data obtained by performing correction similar to that in step S132 on the image data removed in step S131 is acquired as the background color. May be.

  Furthermore, the vehicle length calculation method is not limited to the process shown in FIG. 7 described above, and a sign with a known length and color is attached to the vehicle 12 and imaged by a line sensor, and the vehicle length is calculated based on the image. It may be calculated. FIG. 12 is a diagram illustrating an image obtained by photographing the vehicle 12 to which a sign is attached. That is, when the vehicle 12 to which the circular sign is attached is photographed by the line sensors 31a and 31b or the line sensors 31c and 31d, the circular sign M attached to the vehicle 12 is included as shown in FIG. A dimensional image can be acquired.

  Therefore, as shown in FIG. 7B, the background is removed from the two-dimensional image, and the image of the sign M is extracted. Since the length of the sign M is known, the number of lines (a) in the vehicle traveling direction is obtained from the extracted image of the sign M, and the number of lines (b) in the vehicle traveling direction is obtained from the image of the vehicle. For example, the vehicle length can be calculated by multiplying the length (diameter) and (b / a) of the sign. In the above vehicle length measurement, the number of signs attached to the vehicle 12 is not limited to one, and the vehicle length may be measured by attaching two or more signs to the vehicle 12. In this case, the vehicle length may be obtained by calculating the vehicle length individually based on each sign and calculating the average, or the vehicle length may be calculated after obtaining the average for the number of lines of the sign. Well, various configurations can be employed. In addition, although the number of signs is less time-consuming for preparation, it is possible to calculate the vehicle length more accurately when the number of signs is large.

  Furthermore, in the above-described embodiment, the two line sensors 31a and 31b are attached to the column 20c, and the two line sensors 31c and 31d are attached to the beam 20f. However, it is not essential to attach two line sensors to the column 20c and the beam 20f in order to measure the vehicle length, vehicle height, and vehicle width. Therefore, one line sensor may be used, or three or more line sensors may be used. Of course, increasing the number of line sensors makes it easier to improve vehicle measurement accuracy. However, if the number of line sensors is small, the device configuration is simple and the cost can be reduced. In the above example, the vehicle length is calculated based on the image of the side surface of the vehicle 12, but the vehicle length may be calculated based on the image of the upper surface of the vehicle 12.

It is a perspective view which shows the vehicle measuring device which concerns on one Embodiment of this invention. It is a block diagram which shows the structure of a computer. It is a flowchart which shows the process of a vehicle measurement program. It is a figure which shows the image of a vehicle. It is a flowchart which shows the extraction process of the image of a vehicle. It is a figure which shows the image of a vehicle. It is a flowchart which shows the calculation process of a vehicle length. It is a block diagram which shows the structure of a computer. It is a flowchart which shows the process of a vehicle measurement program. It is a flowchart which shows the removal process of a shadow. It is explanatory drawing explaining the removal process of a shadow. It is a figure which shows the image of a vehicle.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Vehicle measuring device 11 ... Road surface 12 ... Vehicle 20a-20e ... Column 20f ... Beam 21 ... Column 21a, 21b ... Background part 21a2 ... Shadow image 21a3 ... Standard color part 21b2, 21b3 ... Mud dirt 21b4 ... Shadow image 30a , 30b ... entry sensors 30c, 30d ... exit sensors 31a-31d ... line sensor 40 ... computer 40a ... display 40b ... keyboard 40c ... mouse 41a ... sensor control unit 41b ... image extraction unit 41b1 ... reference image setting unit 41b2 ... color difference determination unit 41c ... Vehicle measurement unit 41c1 ... Vehicle length calculation unit 41c2 ... Vehicle height calculation unit 41c3 ... Vehicle width calculation unit 44a ... 1D image data 44b ... 2D image data 44c ... Reference image data 44d ... Vehicle image data 44e ... Inter-sensor distance Data 44f ... Sampling cycle data 44g ... Height line length data 44h Width line length data

Claims (10)

One-dimensional image acquisition means for photographing a predetermined range in a direction perpendicular to the traveling direction of the vehicle traveling on the road surface;
A background portion installed in the photographing range of the one-dimensional image acquisition means;
Two-dimensional image generation means for generating a two-dimensional image by acquiring a vehicle image crossing between the one-dimensional image acquisition means and the background portion a plurality of times by the one-dimensional image acquisition means;
An image that identifies a reference image based on a portion that does not include a vehicle in the two-dimensional image, sequentially compares the reference image with another image, and extracts an image whose difference exceeds a predetermined threshold as a vehicle image Extraction means;
A vehicle measurement apparatus comprising: a measurement unit that measures an outer dimension of the vehicle based on the extracted image. The two-dimensional image generation means acquires an image captured by the one-dimensional image acquisition means within a predetermined time range after passing through the imaging range from before the vehicle passes,
The image extracting means sets a first reference image on the front side of the vehicle in the two-dimensional image and sequentially compares it with an image on the rear side thereof, and sets a second reference image on the rear side of the vehicle. The vehicle measurement device according to claim 1, wherein the vehicle measurement device sequentially compares the image with a front image.   The background included in the photographing range in the background portion is a chromatic color, and the image extraction unit evaluates a difference between the reference image and another image including a hue. The vehicle measurement device according to any one of the above. A vehicle measurement method for measuring an outer dimension of a vehicle based on an image,
A two-dimensional image is generated by performing a one-dimensional image acquisition step of capturing a predetermined range in a direction perpendicular to the traveling direction of the vehicle traveling on the road surface and acquiring an image of the vehicle crossing the front of the background portion a plurality of times. A two-dimensional image generation step;
An image that identifies a reference image based on a portion that does not include a vehicle in the two-dimensional image, sequentially compares the reference image with another image, and extracts an image whose difference exceeds a predetermined threshold as a vehicle image An extraction process;
A vehicle measurement method comprising: a measurement step of measuring an outer dimension of the vehicle based on the extracted image. One-dimensional image acquisition means for photographing a predetermined range in a direction perpendicular to the traveling direction of the vehicle traveling on the road surface;
A background portion installed in the photographing range of the one-dimensional image acquisition means;
A vehicle measurement program for measuring an outer dimension of a vehicle based on an image acquired by an imaging system comprising:
A two-dimensional image generation function for generating a two-dimensional image by acquiring a vehicle image crossing between the one-dimensional image acquisition unit and the background portion a plurality of times by the one-dimensional image acquisition unit;
An image that identifies a reference image based on a portion that does not include a vehicle in the two-dimensional image, sequentially compares the reference image with another image, and extracts an image whose difference exceeds a predetermined threshold as a vehicle image Extraction function;
A vehicle measurement program for causing a computer to realize a measurement function for measuring an outer dimension of a vehicle based on the extracted image. One-dimensional image acquisition means for photographing a predetermined range in a direction perpendicular to the traveling direction of the vehicle traveling on the road surface;
A background portion installed in the photographing range of the one-dimensional image acquisition means;
A background removal means for obtaining an image of a vehicle crossing between the one-dimensional image obtaining means and the background portion by the one-dimensional image obtaining means, and removing the background from the obtained image;
In the image from which the background is removed, a shadow removing unit that removes a shadow that is at the periphery of the shadow of the vehicle and at least outside the end of the image of the vehicle;
A vehicle measuring apparatus comprising: a measuring unit that measures an outer dimension of the vehicle based on an image from which the shadow is removed.   The shadow removing means removes a color having a lower lightness and / or a lower saturation than a color of the background portion, and a color having a higher lightness and / or a higher saturation than a color of a tire. 6. The vehicle measurement device according to 6.   A part of the background part is a reference color part of a known color, and the shadow removing means makes the color of the image of the reference color part become the known color with respect to the image from which the background is removed. The vehicle measurement apparatus according to claim 6, wherein the shadow is removed after correction is performed. A vehicle measurement method for measuring an outer dimension of a vehicle based on an image,
A one-dimensional image acquisition step of capturing a predetermined range in a direction perpendicular to the traveling direction of the vehicle traveling on the road surface and acquiring an image of the vehicle crossing the front of the background;
A background removal step of removing the background from the acquired image;
In the image from which the background is removed, a shadow removing step of removing a shadow that is at the periphery of the shadow of the vehicle and at least outside the edge of the vehicle image;
And a measuring step of measuring an outer dimension of the vehicle based on the image from which the shadow is removed. One-dimensional image acquisition means for photographing a predetermined range in a direction perpendicular to the traveling direction of the vehicle traveling on the road surface;
A background portion installed in the photographing range of the one-dimensional image acquisition means;
A vehicle measurement program for measuring an outer dimension of a vehicle based on an image acquired by an imaging system comprising:
A background removal function for acquiring an image of a vehicle crossing between the one-dimensional image acquisition means and the background portion by the one-dimensional image acquisition means, and removing the background from the acquired image;
In the image from which the background is removed, a shadow removing function that removes a shadow that is at the periphery of the shadow of the vehicle and at least outside the edge of the image of the vehicle;
A vehicle measurement program for causing a computer to realize a measurement function for measuring an outer dimension of a vehicle based on an image from which the shadow is removed.

Images