JP2006091974A - Traffic flow measuring method and device therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To prevent an object not being a vehicle from being recognized as a vehicle in a traffic flow measurement. <P>SOLUTION: It is determined whether or not a vehicle head settlement point, which is selected by subjecting an image picked up with an ITV camera 1 to space differentiation, binarization, masking or the like, is settled based on the result of comparing the peripheral image of the vehicle head settlement point (the image of an area including a point selected as the vehicle head settlement point) with the background image corresponding to the peripheral image. More specifically, when it is determined that the difference between the peripheral image of the vehicle head settlement point and the background image lies in at least a preliminarily fixed level, the vehicle head settlement point is stored, but, if not so, the vehicle head settlement point is deleted from a storage. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a traffic flow measuring method and apparatus, and in particular, from the image information captured by a camera, the presence of a vehicle, the type of vehicle (for example, classification of the size of a vehicle such as a small vehicle, a medium size vehicle), and individual vehicle speed. The present invention relates to a method and apparatus for measuring traffic flow and the like.

  In traffic control systems for general roads and highways, traffic flow is measured by arranging many vehicle detectors on the road side. One of the advanced functions of this measurement is a traffic flow measurement processing system using an ITV (Industrial Television) camera, and research has been conducted in the past. This traffic flow measurement processing system using an ITV camera uses a TV camera as a sensor, and determines the presence and speed of a vehicle by analyzing in real time an image taken by looking down a road from an oblique direction.

For example, in Patent Document 1, in a traffic flow measurement system using an ITV camera, a mask having a width approximately equal to the vehicle width is used after performing spatial differentiation processing on an image output from the ITV camera. A technique for recognizing a vehicle by performing mask processing is disclosed. In such a technique, in an image photographed by a camera, a portion having a different luminance with respect to the background (a portion corresponding to a road in an image photographed by the camera) has been recognized as a vehicle.
JP-A-5-307695

  However, in the conventional techniques including Patent Document 1, a shadow of a leaf or a tall building such as a building that temporarily appears on the road surface is mistakenly recognized as a vehicle, particularly in summer when the sunlight is strong. There was a case. For such a situation where the shadow is mistaken as a vehicle, the use of the time difference of the video of the camera has been studied, but it has not yet been solved.

  The present invention has been conceived in view of the above-described circumstances, and an object of the present invention is to provide a traffic flow measuring method and apparatus capable of avoiding that other than a vehicle is recognized as a vehicle. .

  In the traffic flow measuring method according to the present invention, a step of photographing a road, a plurality of sample points are defined for an image obtained by photographing the road, and luminances of the plurality of sample points are detected. Determining a vehicle head point that is a candidate for a vehicle head position in an image obtained by photographing the road based on the brightness of the plurality of sample points; and information identifying the vehicle head point And an image in the vicinity of the vehicle head point in an image including the vehicle head point, and a background image that is an image in the vicinity of the vehicle head point and does not include a vehicle. Calculating the degree of similarity of the image of the vehicle, and the degree of similarity between the image near the vehicle head point in the image including the vehicle head point and the background image is a predetermined degree or more It is characterized in that it comprises a step of deleting the stored information identifying the headway point.

  According to the present invention, even if an area has a different luminance with respect to a road, an area that is somewhat similar to a background image that includes objects other than vehicles is determined not to be an area that includes vehicles.

  Further, the traffic flow measuring method according to the present invention averages the step of taking a plurality of images of the road in a state where the vehicle is traveling and the plurality of images of the road in a state where the vehicle is traveling. Preferably, the method further includes the step of creating the background image.

  Further, in the traffic flow measuring method according to the present invention, the vicinity of the vehicle head point is an area having a width corresponding to the vehicle with respect to the direction intersecting with the vehicle traveling direction on the road. It is preferable.

  In the traffic flow measuring method according to the present invention, in the step of calculating the degree of similarity of the images, the difference between the image near the vehicle head point in the image including the vehicle head point and the background image is different. In the step of calculating the degree and deleting the storage of the information for identifying the vehicle head point, the image near the vehicle head point in the image including the vehicle head point and the image when the difference is a predetermined value or less It is preferable that the degree of similarity with the background image is greater than or equal to a predetermined degree.

  A traffic flow measuring apparatus according to the present invention defines a plurality of sample points for a camera installed on the side of a road and photographing the road, and an image captured by the camera, and detects the luminance of the plurality of sample points. A vehicle head point determination unit that determines a vehicle head point that is a candidate for a vehicle head position in an image captured by the camera, based on the brightness of the plurality of sample points; and the vehicle head point is specified A vehicle head point storage means for storing information and an image of the vicinity of the vehicle head point in an image including the vehicle head point are images of the vicinity of the vehicle head point created based on an image photographed by the camera. The similarity degree calculating means for calculating the degree of similarity between these images by comparing with a background image that does not include a vehicle, and the similarity degree calculating means includes the above-mentioned image in the image including the vehicle head point. Vehicle head point deletion means for deleting information for identifying the vehicle head point from the storage of the vehicle head point storage means when the degree of similarity between the image in the vicinity of the head point and the background image is a predetermined degree or more. It is characterized by that.

  According to the present invention, even if there is a shadow of a leaf or a shadow of a building such as a building that is temporarily seen on the road, they are also included in the background image. It can avoid being recognized. Therefore, according to this invention, it can avoid recognizing things other than a vehicle as a vehicle.

  Then, by recognizing that something other than the vehicle is recognized as a vehicle, for example, when the vehicle speed is calculated using the position of the vehicle in an image taken by the camera at a certain time interval. If so, the speed of the vehicle is calculated more accurately. In addition, at the intersection, the position of the vehicle can be accurately grasped even when the blue timing of the detection signal is controlled based on the position of the vehicle determined based on the image taken by the camera. Thus, it becomes possible to control the signal in accordance with the actual situation of the road. Furthermore, even when the timing for turning on the right turn signal at the intersection is controlled based on the position of the vehicle determined based on the image captured by the camera, it is possible to control the signal in accordance with the actual situation of the road. Become.

  Hereinafter, an embodiment of a traffic flow measuring device of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating an installation example of the ITV camera 1 in the traffic flow measuring apparatus according to the present embodiment.

  Referring to FIG. 1, an ITV camera 1 is provided on an upper portion of a pole installed on the road side, and a control unit 2 is provided on the lower side of the pole with respect to the ITV camera 1. The field of view (measurement area) of the ITV camera 1 extends over all lanes of a road having four lanes, as indicated by the alternate long and short dash line extending from the ITV camera 1. In FIG. 1, an area α indicated by a two-dot chain line and an area β indicated by a dotted line indicate areas corresponding to image signals output from the ITV camera 1 described later.

  FIG. 2 is a block diagram showing an electrical configuration of the traffic flow measuring apparatus according to the present embodiment.

  Referring to FIG. 2, the traffic flow measuring device mainly includes an ITV camera 1 and a control unit 2. The control unit 2 is based on an image input unit 3 to which an image signal acquired by the ITV camera 1 is input, a vehicle head candidate point detection unit 4, a measurement processing unit 5, and an image signal input to the image input unit 3. And a background image creating unit 7 for creating a road background image. The control unit 2 further includes a transmission unit 6 that transmits information such as traffic flow measurement results calculated by the control unit main body to the traffic control center through a communication line, and power to the control unit 2 from an external power source. Is provided with a power supply unit 8 for controlling the supply of.

  Next, traffic flow measurement processing executed in the control unit main body in the control unit 2 will be described with reference to the flowchart of the traffic flow measurement processing described in FIG. In the traffic flow measuring apparatus of the present embodiment, the timing (interval) at which the traffic flow measuring process as shown in FIG. 3 is started is, for example, the processing capability of the apparatus or an ITV camera handled in a single process. 1 according to the amount of image signal from 1 and the like.

  First, when the image signal output from the ITV camera 1 is captured by the image input unit 3, the vehicle head candidate point detection unit 4 includes the image signal in step S1 (hereinafter, “step” is omitted). Spatial differentiation processing in the traveling direction of the vehicle is executed based on a plurality of predetermined sample points.

  Here, the contents of the spatial differentiation process will be described.

  FIG. 4 is an example of an arrangement of image signals output from the ITV camera 1, and is a diagram schematically showing an arrangement of image signals corresponding to the area indicated by the area α in FIG. In the image input unit 3, as shown in FIG. 4, M sample points are defined along the crossing direction of the road (referred to as the ξ direction) for the image signal output from the ITV camera 1, and the vehicle travels. N sample points along the direction (referred to as η direction) are defined. Thereby, M × N coordinates (i, j) are defined for the image signal. The image input unit 3 stores each luminance value P (i, j) of the M × N coordinates (i, j). If the intervals between the sample points in the η direction and ξ direction are Δη and Δξ, Δη is about 25 cm, and Δξ is about 20 cm, for example. The image input unit 3 can also output the image signal output from the ITV camera 1 to the outside as a video as it is.

  Then, the vehicle head candidate point detection unit 4 uses the luminance values P (i, j) of the M × N coordinates (i, j) stored in the image input unit 3 to perform spatial differentiation processing in the η direction. do. Specifically, the following Sobel operator is applied to each (i, j) (refer to Japanese Laid-Open Patent Publication No. 5-307695 for details of the spatial differentiation process).

  Referring to FIG. 3 again, after the spatial differentiation process in S1, a binarization process is executed in S2. The binarization processing is executed by the vehicle head candidate point detection unit 4 by applying a threshold value Th1 given in advance as a constant and binarizing all pixels subjected to spatial differentiation. The

  Next, in S3, masking is performed.

  The masking candidate point detection unit 4 executes the masking process using a mask prepared in advance according to the classification of a small car, a normal car, a large car, and the like. The contents of the masking process executed in the present embodiment can be the same as the contents described in Japanese Patent Laid-Open No. 5-307695. That is, also in the present embodiment, for example, as shown in FIG. 5, eight types of masks M1 to M8 are prepared and the masking process is executed. M1 to M4 are for ordinary vehicles, and M5 to M8 are for large vehicles. M1, M2, M5, and M6 are masks for two rows, and M3, M4, M7, and M8 are masks for three rows. The pixel of interest in each mask is in the lower left in M1, M3, M5, and M7, and is in the upper left in M2, M4, M6, and M8. In FIG. 5, each mask is hatched at a position corresponding to the pixel of interest.

  In the masking process, the number of pixels having the code 1 existing in each mask is counted, and the counted number is used as a mask score. As an example of a state in which the mask is applied, FIG. 6A shows a state in which the mask M1 is applied in accordance with the pixel of interest (i, j) second from the left and second from the bottom. At this time, since the code 1 exists in the area corresponding to the mask M1, the score is 9. FIG. 6B shows a state in which a mask M2 is applied in accordance with the pixel (i, j) of interest that is second from the left and second from the bottom. The score at this time is 7. In the masking process, the vehicle head candidate point detection unit 4 stores the mask number and score for the pixel of interest in pairs with the coordinates of the pixel. For example, in the example shown in FIG. 6A, information (i, j, M1, 9) is stored for the coordinates (i, j). In the example shown in FIG. 6B, information (i, j, M2, 7) is stored for the coordinates (i, j).

  Referring to FIG. 3 again, it is considered that the vehicle head candidate point detecting unit 4 is photographed in the image currently being processed in S4 using the result of the processing after the masking processing. The vehicle type and vehicle head candidate point of the vehicle are determined. Specifically, the vehicle head candidate point detection unit 4 selects the mask with the highest score as a result of applying eight types of masks (see FIG. 5) to the pixel of interest as described above. If the score of the mask for the large vehicle is the same as the score of the mask for the small vehicle, the mask for the small vehicle is selected. Then, if the score for the selected mask number is equal to or greater than a certain threshold, the vehicle head candidate point detection unit 4 captures the vehicle type corresponding to the mask in the image currently being processed. A possible vehicle type (large vehicle or small vehicle) is set, and the mask is applied again, and the center of gravity is obtained based on the distribution of pixels that is 1, thereby setting the center of gravity as a vehicle head candidate point.

  As a result, the vehicle head candidate point detection unit 4 stores the coordinates of the vehicle head candidate points, the mask number, and the maximum score value in pairs. For example, in the case shown in FIG. 6A, if the coordinates of the center of gravity are (i, j + 5), information (i, j + 5, M1, 9) is stored. In the subsequent processing, the image data and the binarized data are not used, and the processing proceeds based only on the information of the vehicle head candidate point (coordinates of the vehicle head candidate point, mask number, score maximum value).

  Referring to FIG. 3 again, after the process of S4, the vehicle head effective point is extracted in S5 in the control unit body. Specifically, the extraction of the vehicle head effective point means that the measurement processing unit 5 sequentially examines a plurality of vehicle head candidate points stored in the vehicle head candidate point detection unit 4 to find the most likely vehicle head position (vehicle head effective point). ) To extract. More specifically, in the extraction of the effective point of the vehicle head according to the present embodiment, as described in JP-A-5-307695 as the extraction of the effective point of the vehicle head, the measurement processing unit 5 has a region close to the vehicle (for example, approximately When there are n vehicle head candidate points in an area where one vehicle exists, the first (n = 1) vehicle head candidate points are first read from the vehicle head candidate point detection unit 4 and registered as vehicle head effective points. Next, the measurement processing unit 5 sequentially reads the score (maximum score value) of the vehicle head candidate points after n = 2 from the vehicle head candidate point detection unit 4 and compares it with the score of the vehicle head effective point of n = 1. The larger one of the two is newly registered as a vehicle head effective point, or a point close to the traveling direction of the vehicle is newly registered as a vehicle head effective point. When a vehicle head candidate point that has not become a vehicle head effective point is deleted from registration in the vehicle head candidate point detection unit 4 and a vehicle head candidate point is newly registered as a vehicle head effective point, it is compared with the vehicle head candidate point. Delete the registration of the valid vehicle head. In this way, one vehicle head effective point is determined from a plurality of vehicle head candidate points that are close to each other.

  After the process of S5, in S6, it is judged by the measurement process part 5 whether the vehicle head effective point was obtained, and a process is complete | finished when not obtained. On the other hand, when the vehicle head effective point is obtained, a vehicle head fixed point is selected from the vehicle head effective points in S7.

  The selection of the vehicle head definite point means that the measurement processing unit 5 determines the position of the vehicle head effective point registered in the measurement processing unit 5 as described in Japanese Patent Application Laid-Open No. 5-307695 as definite vehicle head definite point. If there are m vehicle head effective points, the first vehicle head effective point is registered as the vehicle head definite point, then the next vehicle head effective point and the vehicle head effective point registered as the vehicle head definite point From the positional relationship between the two, it is determined whether or not both points are within the range of the length and width of a vehicle such as a large vehicle or a small vehicle corresponding to the mask. If it is determined that the vehicle is within the range, the vehicle head determination point and the next vehicle head effective point that are currently compared are selected as the vehicle head determination point, and the other points are determined as the vehicle head determination point. Delete from point candidates. In this way, each vehicle head effective point registered in the measurement processing unit 5 is examined, and the remaining vehicle head effective points are selected and stored as vehicle head fixed points.

  Next, in the control unit main body, a background image is created by the background image creation unit 7 in S8. The background image creation unit 7 creates a background image based on an image signal input from the ITV camera 1 via the image input unit 3 as shown in FIG. Specifically, the background image creation unit 7 averages a plurality of image signals generated when the ITV camera 1 captures a plurality of states of a road on which the vehicle travels, thereby obtaining a background image for the road. Create an image that does not include the vehicle.

  Next, in S9, the measurement processor 5 calculates the difference A between the image of the region including the point selected as the vehicle head definite point and the background image created corresponding to the vehicle definite point. The

  Here, the “near” region of the vehicle head definite point will be described with reference to FIG. FIG. 7 is an example of an arrangement of image signals output from the ITV camera of FIG. 1, and is a diagram schematically showing an arrangement of image signals for a region in the vicinity of the vehicle head definite point.

  In FIG. 7, sample points selected as vehicle head definite points are described as “x”. The image signal shown in FIG. 7 includes the vehicle head definite point at the top in the vehicle traveling direction and the center in the vehicle width direction, and approximately 10 times Δξ in the vehicle width direction. Includes sample points that exist within a rectangular region of about 10 to 20 times Δη. If the vehicle fixed point in FIG. 7 corresponds to the central part of the bumper of the vehicle, the rectangular area as shown in FIG. 7 is often enough to detect the entire surface of the windshield from the bumper of the vehicle. A region having an area (a region corresponding to the region β in FIG. 1) is considered.

Further, the measurement processing unit 5 calculates the degree of difference A based on the following equation (1). In Equation (1), K is the number of sample points included in a region (see FIG. 7) including the points selected as the vehicle head fixed point, and X _i is the point selected as the vehicle head fixed point. This is the luminance of the i-th sample point in the image of the included region, and B _i is the luminance of the i-th sample point in the background image (see FIG. 4).

  When the difference A is calculated in S9, the calculated difference A is compared with a predetermined value L by the measurement processing unit 5 in S10. The predetermined value L is assumed to be stored in a predetermined memory in the measurement processing unit 5 in advance. Then, if it is determined that A is greater than L, that is, it is determined that the image of the area including the point selected as the vehicle head decision point and the background image are different from each other beyond a predetermined level. If so, the process proceeds to S11. On the other hand, if A is L or less, the process proceeds to S12.

  In S11, it is assumed that the vehicle head fixed point which is the current processing target has been fixed, and is stored in the measurement processing unit 5 by the measurement processing unit 5, and the process proceeds to S13. On the other hand, in S12, the vehicle head definite point currently being processed is deleted from the measurement processing unit 5 by the measurement processing unit 5, and the process proceeds to S13.

  In S13, a determination is made as to whether or not the processing of S8 to S11 has been performed for all vehicle head fixed points stored in the measurement processing unit 5, and there are still vehicle head fixed points that have not been processed in S11. If it is determined that there is, the process returns to S8. On the other hand, if it is determined that the process for determining all the vehicle head determined points in S11 has been performed, the process proceeds to S14.

  Through the processes of S1 to S13 described above, the control unit main body has obtained a point (vehicle head definite point) corresponding to the head of the vehicle in the image for a certain frame captured by the ITV camera 1.

  In S14, the measurement processing unit 5 calculates the speed of the vehicle using the vehicle head fixed point determined in the measurement processing unit 5, and then the process proceeds to S15. Specifically, the vehicle speed is determined by examining the positional relationship between the vehicle head fixed point in the measurement area of the currently processed frame and the vehicle head fixed point found in the measurement area one frame before. Calculate the speed of Note that the well-known technique described in Japanese Patent Laid-Open No. 5-307695 can also be applied to the calculation of the vehicle speed.

  In S15, the vehicle processing point 5 determined by the measurement processing unit 5 (position information thereof), the vehicle type (large vehicle or small vehicle) of the vehicle corresponding to the vehicle head determination point, and the vehicle Is output to an external device via the transmission unit 6 as a result of the traffic flow measurement, and the traffic flow measurement process is terminated.

  In the present embodiment described above, a vehicle head definite point selected by spatial differentiation, binarization, masking, etc. with respect to an image taken by the ITV camera 1 is further displayed as a peripheral image of the vehicle definite point (vehicle head). The image of the region including the point selected as the fixed point) is compared with the background image, and it is determined whether or not to be fixed based on the comparison result. Specifically, if it is determined that the peripheral image of the vehicle head definite point and the background image are different from each other beyond a predetermined level, the vehicle definite point is stored in the measurement processing unit 5 as the vehicle definite point. However, otherwise, the vehicle head definite point is deleted from the storage of the measurement processing unit 5. By such processing, a shadow of a leaf of the tree is generated on the road, and even if a sample point corresponding to the shadow of the leaf of the tree is erroneously selected in S7 as the vehicle head definite point in the image taken by the ITV camera 1, such a vehicle head The fixed point can be deleted in the processing from S8 to S12. This is because such a shadow of a leaf is considered to be included in the background image. That is, when such an image including a leaf shadow is compared with a background image, it is considered that these images do not differ beyond a predetermined level. Therefore, the above value L should be determined in consideration of such points.

  Further, in the present embodiment, as a process for determining whether or not the difference exceeds a predetermined degree as described above, the difference A is calculated and the value is determined. Such processing is not limited to calculation of the degree of difference. For example, the degree of similarity (for example, normalized correlation) is obtained for the surrounding image and background image of the vehicle head fixed point, and if the similarity is less than a predetermined value, the surrounding image and background image of the vehicle head fixed point are determined in advance. It may be determined that the difference exceeds the specified level.

  In the present embodiment, when it is not determined that the peripheral image of the vehicle head fixed point and the background image are different from each other beyond a predetermined level, the vehicle head fixed point is deleted. That is, “when the degree of difference between the surrounding image of the vehicle head definite point and the background image is equal to or less than a predetermined level” is “the image near the vehicle head point in the image including the vehicle head point and the background image. Corresponds to a case where the degree of similarity is greater than or equal to a predetermined degree.

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

It is a figure which shows the example of installation of the ITV camera in the traffic flow measuring device which is one embodiment of this invention. It is a figure which shows the electric constitution of the traffic flow measuring apparatus of FIG. It is a flowchart of the traffic flow measurement process performed with the control part main body of FIG. It is a figure which shows typically an example of the arrangement | sequence of the image signal output from the ITV camera of FIG. It is a figure which shows typically eight types of masks prepared for the vehicle head candidate point detection part of FIG. (A) is a figure for demonstrating the state by which mask M1 was applied to the attention pixel in the masking process of FIG. 3, (B) is for demonstrating the state by which mask M2 was applied to the attention pixel. FIG. FIG. 2 is an example of an array of image signals output from the ITV camera of FIG. 1, and is a diagram schematically illustrating an array of image signals for a region in the vicinity of a vehicle head definite point.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ITV camera, 2 control part, 3 image input part, 4 vehicle head candidate point detection part, 5 measurement process part, 6 transmission part, 7 background image creation part, 8 power supply part.

Claims (5)

Shooting the road,
Defining a plurality of sample points for an image obtained by photographing the road, and detecting the luminance of the plurality of sample points;
Determining a vehicle head point that is a candidate vehicle head position in an image obtained by photographing the road based on the brightness of the plurality of sample points;
Storing information identifying the vehicle head point;
The degree of similarity between these images by comparing an image in the vicinity of the vehicle head point in the image including the vehicle head point with a background image that is an image in the vicinity of the vehicle head point and does not include a vehicle. Calculating steps,
If the degree of similarity between the image near the vehicle head point in the image including the vehicle head point and the background image is greater than or equal to a predetermined level, deleting the information specifying the vehicle head point; and A traffic flow measuring method comprising: Capturing a plurality of images of the road in a state in which the vehicle is traveling;
The traffic flow measuring method according to claim 1, further comprising: creating the background image by superimposing and averaging a plurality of images of the road in a state where the vehicle is traveling.   3. The traffic flow according to claim 1, wherein the vicinity of the vehicle head point is a region having a width corresponding to a vehicle with respect to a direction intersecting with a traveling direction of the vehicle on the road centered on the vehicle head point. Measurement method. In the step of calculating the degree of similarity of the images, the degree of difference between the image near the vehicle head point in the image including the vehicle head point and the background image is calculated,
In the step of deleting the storage of the information specifying the vehicle head point, when the difference is a predetermined value or less, the similarity between the image in the vicinity of the vehicle head point in the image including the vehicle head point and the background image The traffic flow measuring method according to any one of claims 1 to 3, wherein a degree to be performed is greater than or equal to a predetermined degree. A camera installed on the side of the road and shooting the road;
A plurality of sample points for an image captured by the camera, and a luminance detecting means for detecting the luminance of the plurality of sample points;
Vehicle head point determination means for determining a vehicle head point, which is a candidate point of the vehicle head position in the image captured by the camera, based on the luminance of the plurality of sample points;
Vehicle head point storage means for storing information identifying the vehicle head point;
A background image that is an image in the vicinity of the vehicle head point that is created based on an image photographed by the camera, and is an image that does not include a vehicle, in the image including the vehicle head point. A similarity degree calculating means for calculating the degree of similarity between these images by comparing with
If the degree of similarity between the image near the vehicle head point in the image including the vehicle head point and the background image is greater than or equal to a predetermined degree, the similarity degree calculating means includes information for specifying the vehicle head point. A traffic flow measuring device comprising: a vehicle head point deletion unit that deletes the vehicle head point storage unit from the storage.

Images