JP4282158B2 - Vehicle end point detection method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a vehicle specification automatic measurement system for measuring the dimensions of a vehicle traveling on a road, and more particularly to a method for detecting front and rear and side end points of a vehicle.
[0002]
[Prior art]
The Ministry of Construction and the National Police Agency regulate the passage of special vehicles to preserve roads and prevent danger. At the time of this control, a special vehicle is guided into the control base and its specifications (vehicle length, vehicle height, vehicle width) are measured. Conventionally, features such as the contour of the vehicle are detected from the image of the vehicle captured by the CCD camera by processing such as spatial differentiation, and the position of the vehicle end point is detected by detecting the position of the end point of the vehicle from the arrangement and shape of the features. I was measuring the width.
[0003]
[Problems to be solved by the invention]
However, the conventional method can reliably detect the end points of the vehicle if the vehicle itself and surrounding buildings are not affected by shadows, and if the vehicle has a simple box shape. However, the end point of the vehicle body may not be detected when a shadow is generated or when the vehicle body has a complicated shape. Further, even when the ground and the vehicle body are almost the same color, it is difficult to detect the contour of the vehicle body, and thus it may be difficult to detect the end points of the vehicle body.
[0004]
The present invention solves such a conventional problem, and has an object to accurately detect an end point of a vehicle body even under a weather condition in which a vehicle having a complicated shape or a shadow is generated.
[0005]
[Means for Solving the Problems]
In order to solve the above problems, the present invention detects an end point of the vehicle by imaging a vehicle existing on the road by an imaging device installed on the road and processing image data output from the imaging device. In this method, a measurement pattern for generating image data having a plurality of luminance levels is formed on the road surface, and image data output from the imaging device in a state where the vehicle is present on the measurement pattern. The position of the end point of the vehicle is detected by processing and specifying the position where the measurement pattern cannot be detected. With this configuration, it is possible to accurately detect the end points of the vehicle body and accurately measure the vehicle dimensions even under weather conditions where a vehicle having a complicated shape or a shadow is generated.
[0006]
DETAILED DESCRIPTION OF THE INVENTION
According to a first aspect of the present invention, an end point of the vehicle is obtained by imaging a vehicle existing on the road by an imaging device installed on the road and processing image data output from the imaging device. In the detection method, a measurement pattern for generating image data having a plurality of luminance levels is formed on a road surface, and the output from the imaging device in a state where the vehicle exists on the measurement pattern A method for detecting the position of an end point of the vehicle by processing image data and specifying a position where the measurement pattern cannot be detected , and includes three black and white belt-like patterns arranged adjacent to each other. A line segment that configures the measurement pattern and intersects the three strip-shaped patterns on the image captured by the imaging device is set. Calculate the differential intensity at the boundary between one white part and the black part of the measurement pattern and the differential intensity at the other white part and black part, and the sum of the two obtained differential intensities is greater than or equal to a preset threshold value. And a vehicle end point detection method for determining that a measurement pattern can be detected on the line segment when a difference between two obtained differential intensities is equal to or less than a preset threshold value , and is formed on a road surface. In addition to detecting the position of the end point of the vehicle by specifying the position where the measured pattern cannot be detected , each of the sum and difference of the differential intensities at the boundary between the two white portions and one black portion calculated on the line segment Is compared with a preset threshold value to determine whether pattern detection is possible or not .
[0014]
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0015]
FIG. 1 is a diagram showing an overall configuration of an embodiment of the present invention. In FIG. 1, imaging devices 1a to 1d arranged in a positional relationship corresponding to rectangular vertices at four locations above the vehicle are devices that capture the vehicle and generate image data. The imaging devices 1a to 1d are connected to an image processing device (not shown). The image processing apparatus measures the position of the end point on the vehicle head, the vehicle rear, and the vehicle side by performing predetermined image processing on the input image data. The vehicle length measuring pattern 2 is a pattern formed on the road surface along the traveling direction of the vehicle, and the vehicle length measuring pattern 2 is detected along the traveling direction of the vehicle on the image captured by the imaging device. This is for detecting the position of the vehicle head / vehicle tail by specifying the position where the pattern cannot be detected. The vehicle width measurement pattern 3 is a pattern formed on the road surface at a right angle to the traveling direction of the vehicle. On the image captured by the imaging device, the vehicle width measurement pattern 3 is perpendicular to the traveling direction of the vehicle. The position of the vehicle side is detected by identifying the position where the pattern cannot be detected.
[0016]
FIG. 2 is a diagram showing an example of a pattern applied to the road surface in the embodiment of the present invention. Each of the vehicle length measurement pattern 2 and the vehicle width measurement pattern 3 is composed of a belt-like pattern in which the center is black and both sides are white. The vehicle length measurement pattern 2 is provided in parallel with the traveling direction of the vehicle, and the vehicle width measurement pattern 3 is disposed perpendicular to the traveling direction of the vehicle. The width of the vehicle length measurement pattern 2 is narrower than the width of the vehicle to be measured. The vehicle width measurement pattern 3 is shorter than the length of the vehicle to be measured.
[0017]
In addition, although the example of the simplest pattern was shown in FIG. 2, according to the measurement range of the vehicle length and the vehicle width to be measured, the vehicle length measurement pattern has a vehicle head measurement portion and a vehicle tail measurement portion. It may be separated into parts, or a plurality of vehicle width measurement patterns may be arranged in the vehicle traveling direction. Further, although the white and black patterns are shown in FIG. 2, any patterns may be used as long as they have different luminance levels in the image data output from the imaging apparatus.
[0018]
FIG. 3 is a diagram showing details and features of a pattern portion of an image taken by the imaging apparatus. The pattern has a pattern in which two white parts composed of a white part # 1 and a white part # 2 sandwich a black part. Therefore, it has a boundary # 1 that is a boundary between the white part # 1 and the black part, and a boundary # 2 that is a boundary between the black part and the white part # 2.
[0019]
In this embodiment, a certain line segment orthogonal to boundary # 1 and boundary # 2 is defined as a pattern detection line segment, and a pattern detection method for this pattern detection line segment is disclosed. Since the positions of the boundary # 1 and the boundary # 2 in a certain pattern detection line segment on the images captured by the imaging devices 1a to 1d can be set when the imaging devices 1a to 1d are installed, the boundary # 1 registration position and boundary # 2 registration position are set. The distance between the boundary # 1 registration position and the boundary # 2 registration position is defined as a boundary registration width. Further, when the spatial differential intensity is calculated for the captured image, the spatial differential intensity at the boundary # 1 registration position is the boundary # 1 intensity, and the spatial differential intensity at the boundary # 2 registration position is the boundary # 2 intensity. Also, in the actually captured image, the differential intensity does not necessarily become maximum at the registered position, such as when the vehicle body is on a certain pattern detection line segment, so the spatial differential intensity maximum near the boundary # 1 registered position. Is the boundary # 1 peak position, the maximum value of the spatial differential intensity is the boundary # 1 peak intensity, the position of the spatial fraction maximum near the boundary # 2 registered position is the boundary # 2 peak position, the spatial differential intensity maximum The value is the boundary # 2 peak intensity, and the distance between the boundary # 1 peak position and the boundary # 2 peak position is the boundary peak width.
[0020]
In the present embodiment, this pattern detection line segment is virtually moved to the left and right of the drawing, that is, in parallel with the longitudinal direction of the pattern, and the pattern detection is sequentially attempted at the position of the pattern detection line segment. The position where it cannot be made is the detection position of the end point of the vehicle.
[0021]
3 shows the case of the vehicle length measurement pattern 2 in FIG. 2, but the same is true for the vehicle width measurement 3 only by rotating the figure by 90 degrees. .
[0022]
FIG. 4 is a diagram for explaining a first example of the pattern detection method. In this method, it is determined whether pattern detection is possible or not based on the differential intensity at the boundary between the white part and black part of the pattern. That is, paying attention to a certain pattern detection line segment in the image captured by the imaging device, spatial differentiation processing is performed on the pattern detection line segment, and the boundary # 1 intensity and boundary # 2 intensity and the luminance value of the black portion are calculated. To do. When the boundary # 1 or boundary # 2 intensity is larger than the differential intensity preset for each luminance value of the black portion, it is determined that the pattern can be detected at the position of the pattern detection line segment. It is assumed that the pattern cannot be detected when the differential intensity of is low. Here, the differential intensities necessary to enable pattern detection are set for each black portion luminance value. Generally, when the black portion has a high luminance value, the image is clearer and darker. This is because the differential strength tends to increase. Therefore, the differential intensity necessary for enabling the pattern detection for each black portion luminance value can be expressed by an appropriate function without necessarily setting for each black portion luminance value.
[0023]
FIG. 5 is a diagram for explaining a second example of the pattern detection method. In this method, it is determined whether pattern detection is possible or not based on the sum and difference of differential intensities at two boundaries between the pattern white portion and the black portion. That is, paying attention to a certain pattern detection line segment in the image captured by the imaging device, spatial differentiation processing is performed on the pattern detection line segment. Boundary # 1 and boundary # 2 intensities are calculated. When the sum of boundary # 1 intensity and boundary # 2 intensity is greater than or equal to a preset threshold value and the difference between the two obtained differential intensities is less than or equal to a preset threshold value, that is, boundary # 1 intensity and Only when the boundary # 2 intensity relationship falls within the gray region of the graph of FIG. 5, it is assumed that the pattern can be detected at the position of the pattern detection line segment, and otherwise, the pattern cannot be detected. Here, the upper limit is set for the difference between the boundary # 1 intensity and the boundary # 2 intensity. If the position of the focused pattern detection line segment is really a pattern, the boundary # 1 intensity and the boundary # 2 intensity are substantially equal. This is because it is expected to become a value.
[0024]
FIG. 6 is a diagram for explaining a third example of the pattern detection method. In this method, it is determined whether pattern detection is possible or not based on the width between two borders between the pattern white part and the black part. That is, paying attention to a certain pattern detection line segment in the image picked up by the imaging device, spatial differentiation processing is performed on the pattern detection line segment, and the differential peak positions in the vicinity of boundary # 1 and boundary # 2 are calculated. The boundary peak width is calculated from the distance between peak positions. And, when the difference between the obtained boundary peak width and the boundary registration width registered in advance is smaller than a certain threshold value, pattern detection is possible, and in other cases, pattern detection is impossible. To do.
[0025]
FIG. 7 is a diagram for explaining a fourth example of the pattern detection method. In this method, it is determined whether pattern detection is possible or not based on the luminance values of the pattern white part and black part. That is, in the image picked up by the image pickup device, paying attention to a certain pattern detection line segment, the brightness values of the pattern white part and the black part are calculated. It is assumed that the pattern can be detected when the luminance value of the pattern white portion is larger than a threshold value registered in advance, and the pattern cannot be detected in other cases. Further, it is assumed that the pattern can be detected when the luminance value of the pattern black portion is smaller than a threshold value registered in advance, and the pattern cannot be detected in other cases.
[0026]
FIG. 8 is a diagram for explaining a fifth example of the pattern detection method. In this method, it is determined whether pattern detection is possible or not based on the ratio between the white portion luminance value and the black portion luminance value of the pattern. That is, in the image picked up by the image pickup device, paying attention to a certain pattern detection line segment, the brightness values of the pattern white part and the black part are calculated. It is assumed that the pattern can be detected when the ratio of the white portion luminance value to the black portion luminance value is larger than a threshold value registered in advance, and in other cases, the pattern detection is impossible.
[0027]
FIG. 9 is a diagram for explaining a sixth example of the pattern detection method. In this method, it is determined whether pattern detection is possible or not by calculating the correlation of the luminance change on the line segment corresponding to the same line on the road surface on the two images captured by the two imaging devices. . The right image and the left image in FIG. 9 are arranged in a direction orthogonal to the pattern like the imaging device 1a and the imaging device 1b or the imaging device 1c and the imaging device 1d with respect to the vehicle length measurement pattern 2 in FIG. 2 are images captured by two imaging devices. Corresponding pattern detection line segments correspond to the same position on the road surface. The position of the pattern detection line segment in the right image and the position of the pattern detection line segment in the left image indicating the same position on the road surface are registered in advance when the imaging apparatus is installed. The right image luminance value sequence and the left image luminance value sequence are obtained by extracting the luminance values on the left and right pattern detection line segments as a one-dimensional luminance value sequence.
[0028]
Then, the correlation coefficient between the right image luminance value sequence and the left image luminance value sequence is calculated, and the pattern can be detected when the correlation coefficient value is larger than the threshold value registered in advance, that is, when the left and right correlation is high. In other cases, it is assumed that the pattern cannot be detected. This is because when the pattern is actually seen at the positions of the corresponding pattern detection line segments on the left and right, the left and right luminance value sequences have almost the same shape, so it can be expected to show a high correlation. If the position of the minute is hidden by the vehicle body, the left and right pattern value line segments are visible at the positions of the left and right pattern detection line segments due to the parallax of the left and right imaging devices. This is because of a low.
[0029]
Here, when the angle of view is sufficiently adjusted so that the pattern widths on the images captured by the left and right imaging devices are almost equal, the right image luminance value sequence and the left image luminance value sequence are The pixel value column on the pattern detection line segment may be used, but if the angle of view of the imaging device is not sufficiently adjusted and the widths of the patterns on the left and right images are different, the right image luminance value It is necessary to perform interpolation between pixels so that the lengths of the columns and the left image luminance value columns are equal.
[0030]
Note that the left and right luminance value sequences can be compared not by the correlation coefficient but by the difference, but the brightness of the left and right images is not necessarily the same due to the aperture control of the left and right imaging devices and the positional relationship with the sun. In general, the correlation coefficient can be compared more stably than the difference.
[0031]
Although six types of pattern detection methods have been described above, it is of course possible to perform pattern detection under an AND condition or an OR condition by combining all or some of these detection methods.
[0032]
【The invention's effect】
As described above, in the present invention, a measurement pattern for generating image data having a plurality of luminance levels is formed on the road surface, and is output from the imaging device in a state where a vehicle exists on the measurement pattern. Since the position of the end point of the vehicle is detected by processing the image data and specifying the position where the measurement pattern cannot be detected, the vehicle has a complicated shape or the weather conditions in which a shadow is generated However, there is an effect that the end point of the vehicle body can be accurately detected and the vehicle dimensions can be accurately measured.
[Brief description of the drawings]
FIG. 1 is a diagram showing an overall configuration of an embodiment of the present invention;
FIG. 2 is a diagram showing an example of a pattern according to the embodiment of the present invention;
FIG. 3 is an explanatory diagram of details and features of a pattern of an image captured by the imaging device;
FIG. 4 is an explanatory diagram of a first example of a pattern detection method;
FIG. 5 is an explanatory diagram of a second example of a pattern detection method;
FIG. 6 is an explanatory diagram of a third example of a pattern detection method;
FIG. 7 is an explanatory diagram of a fourth example of a pattern detection method;
FIG. 8 is an explanatory diagram of a fifth example of the pattern detection method;
FIG. 9 is an explanatory diagram of a sixth example of the pattern detection method.
[Explanation of symbols]
1a to 1d Imaging device 2 Vehicle length measurement pattern 3 Vehicle width measurement pattern

Claims (1)

An image having a plurality of luminance levels in a method of detecting an end point of the vehicle by imaging a vehicle existing on the road by an imaging device installed on the road and processing image data output from the imaging device A measurement pattern for generating data is formed on the road surface, the image data output from the imaging device is processed in a state where the vehicle is present on the measurement pattern, and the measurement pattern is detected. A method for detecting a position of an end point of the vehicle by specifying a position where it cannot be performed , wherein the measurement pattern is configured by three black and white belt-like patterns arranged adjacent to each other, and the imaging device A line segment intersecting with the three strip-shaped patterns is set on the captured image, and one white portion and a black color of the measurement pattern are set on the line segment. The differential strength of the boundary of the white and the differential strength of the other white portion and the black portion are respectively calculated, and the sum of the two obtained differential strengths is not less than a preset threshold value, and the two obtained differential strengths A vehicle end point detection method characterized in that it is determined that a measurement pattern can be detected on the line segment when the difference between the two is equal to or less than a preset threshold value .

Images