JP3742410B2 - Traffic flow monitoring system for moving objects - Google Patents

Description

The present invention relates to a traffic flow monitoring system for a moving object that monitors how a moving object such as a vehicle moves along a restricted route such as a road.
For example, a traffic flow monitoring system that monitors the movement of vehicles on the road detects various information such as the type, quantity, and average speed of moving objects moving on the road, displays road conditions, predicts traffic jams, and It is useful for various purposes such as control of ventilating equipment.

Conventionally, information such as the type, quantity, and average speed of moving objects described above has been obtained by dedicated sensors installed at predetermined locations, but each of these sensors has a fixed role. If another information is needed, another sensor must be installed.
Taking these points into consideration, instead of these dedicated sensors, surveillance cameras were installed near the interchange, near the entrance / exit of the tunnel, and inside the tunnel, and captured by these cameras. A system for extracting the above-described various information from video has been proposed, and great expectations are placed on its high expandability.

FIG. 7 shows a configuration example of a conventional traffic flow monitoring system for moving objects.
In FIG. 7, the video captured by the video camera 301 installed at the sample position is input to the difference processing unit 303 via the analog-digital conversion processing unit (A / D) 302 and held in the still image memory 304. The difference from the background image is obtained.
In response to the difference input for each pixel obtained by the difference processing unit 303, the binarization processing unit 305 discriminates between a pixel having a difference from the background image and a pixel having no difference based on a predetermined threshold. A binarized image indicating the discrimination result is created and sent to the detection processing unit 306.

The detection processing unit 306 is configured to detect an image having a predetermined feature from the input binarized image of each frame as an image representing a vehicle, and use the position for the processing of the identification processing unit 307. .
At this time, the detection processing unit 306 particularly sets a part near the side where the video of the vehicle enters the binarized image as a detection area for detecting a new vehicle, and an image representing the vehicle appears in this detection area. Sometimes it is detected as a new vehicle and sent to the counting processing unit 308.

In FIG. 7, the count processing unit 308 counts the number of vehicles that have passed through the field of view of the camera 301 at a predetermined time based on the detection result by the detection processing unit 306 described above. It is configured to notify.
Further, the identification processing unit 307 associates the vehicle detected in the current frame with the vehicle detected in the previous frame based on the detection result obtained by the detection processing unit 306 for each frame. The combination of the position of the attached vehicle in the current frame and the previous frame is used as a result of the identification process, and is provided to the process of the speed calculation unit 309.

For example, the identification processing unit 307 discriminates the closest vehicle from the detection result for the current frame in a predetermined range on the traveling direction side of the vehicle relative to the position of the vehicle in the previous frame. What is necessary is just to send out the combination of a position and the position of the vehicle in the said frame as a corresponding vehicle identification process result.
In response to the input of the identification processing result, the speed calculation unit 309 first, for each vehicle, the corresponding vehicle at the time corresponding to each frame from the position difference on the screen of the corresponding vehicle and the time between the frames. Calculate the speed on the road. Next, the speed calculation unit 309 obtains an average value of speeds on the road corresponding to these frames obtained for each vehicle, and calculates the average value of the corresponding vehicle in the section corresponding to the field of view of the camera 301. The center 310 is notified.

Here, in a normal traffic flow monitoring system, the camera 301 is installed so that the optical axis is inclined with respect to the road so that a wide range can be monitored by one camera 301.
Therefore, in the speed calculation unit 309, it is necessary to convert the length in the binarized image into the length of the section on the road corresponding to the field of view of the camera 301. It is convenient if the length of the corresponding section on the road is known.

The present applicant has already filed Japanese Patent Application No. 6-47452 “Position Association Method, Position Association Device, Vehicle Speed Calculation Method, and Vehicle Speed Calculation Device” as a technique for simplifying such conversion processing. is doing.
This technique is based on the size and spacing of an image representing a marker with a known size and spacing, such as a white line on the road, and the actual length of the section on the road corresponding to each pixel on the screen. Is used for specifying the position of the moving object and calculating the speed.

If this technique is used, based on the image showing the white line on the road shown in FIG. 8A, as shown in FIG. 8B, the position of each pixel on the binarized image (in FIG. 8). And the position on the road can be obtained, and the speed of the vehicle can be easily obtained by using a function indicating the correspondence.
In addition, the center 310 collects information on the number and speed of vehicles obtained by the traffic flow monitoring system provided in various places on the road, and performs traffic flow analysis processing based on such information, thereby causing traffic congestion and the like. It is possible to grasp the road conditions, predict the occurrence and resolution of traffic jams, and discover accidents. Thereby, accurate information is provided to the user, and the service to the user is improved.

In addition, application services such as displaying a warning message for speeding vehicles based on the measured speed values obtained for individual vehicles are also provided.
Japanese Patent Application No. 6-47452

By the way, in order to perform the above-described processing on all the frames captured by the camera 301, a huge amount of calculation is required. For this reason, in a conventional traffic flow monitoring system, sample frames are extracted at predetermined time intervals, and the number of moving objects is calculated and the speed is calculated using the extracted sample frames, thereby reducing the amount of calculation. It was.
In this case, it is necessary to determine the detection area for detecting a new vehicle and the size of the range on the traveling direction side used for vehicle identification in consideration of the sample frame extraction interval and the vehicle speed.

However, in the conventional traffic flow monitoring system, the size of the detection area and the range used for vehicle identification is left to the subjective judgment of the user, and the detection area and the range used for identification are set large. In many cases, it was too small or too narrow.
For example, if the detection area is set too large, a new vehicle that has entered the field of view of the camera 301 can be captured without omission, but an already detected vehicle is recognized as a new vehicle again. Since there is a possibility, the amount of calculation required for the detection process becomes enormous.

On the other hand, if the setting of the detection area is too narrow, the amount of calculation can be reduced, but there is a possibility that detection of a vehicle that has newly entered the field of view may fail.
Also, in the conventional traffic flow monitoring system, as described above, the vehicle identification process is simply performed in a predetermined area ahead of the traveling direction on the assumption that the vehicle speed and the traveling direction do not change greatly. A vehicle that is closest to the vehicle position in the previous frame is selected and identified.

However, when a vehicle actually travels on a road, sudden acceleration, deceleration, or direction change frequently occurs due to lane change or overtaking, and the above-mentioned method does not always give an accurate identification result. I can't say I can.
For example, when the vehicle b continuously travels behind the vehicle a and the vehicle c traveling in the adjacent lane has interrupted between the vehicle a and the vehicle b, the front It can be considered that the vehicle c is mistakenly identified as the vehicle b in the current frame because of the rapid speed change between the vehicle b and the vehicle c generated between the current sample frame and the current sample frame. If such a misidentification occurs, the subsequent speed calculation process cannot be performed accurately.

Further, in the conventional traffic flow calculation system, as described above, in the calculation process of the average speed of the vehicle, the speed in each sample frame obtained from the difference in vehicle position between each sample frame and the previous sample frame is calculated. Simply average.
On the other hand, as can be seen from the curve shown by the solid line in FIG. 8B, the slope of the curve is larger in the upper part of the binarized image corresponding to the part far from the camera 301 than in the lower part. The distance on the road corresponding to one pixel is large. That is, as shown in FIG. 8 (b), the resolution at each position of the binarized image decreases monotonously with a downwardly convex curve as the position on the screen approaches from the lower side to the upper side. I understand.

Therefore, when the traffic flow is monitored using the camera 301 disposed obliquely with respect to the road as described above, the position detection accuracy decreases as the vehicle moves away from the camera 301 and the image approaches the upper side of the screen. I understand that
However, in the conventional method, the position detection results with different precisions are handled equally, and as a result, the measurement accuracy of the vehicle speed as a whole has been lowered.

As described above, there are various problems in the conventional traffic flow monitoring system using image information, but images of each frame including a huge amount of information can be obtained continuously and various information can be extracted. Because it can, it has excellent extensibility.
For example, if an object that is stationary over a plurality of consecutive sample frames is extracted, it is possible to obtain information on the presence or absence of an object that obstructs the traveling of the vehicle, such as a falling object, and regardless of the speed of the vehicle, Various information such as the inter-vehicle distance can be obtained for this vehicle.

Therefore, there is a need for a method for automatically obtaining accurate information from image information in order to utilize the characteristics of the traffic flow monitoring system using image information and provide users with more detailed and accurate information. Yes.
An object of the present invention is to provide a traffic flow monitoring system that makes it possible to calculate the quantity and speed of moving objects more precisely by using the characteristics of image information.

FIG. 1 is a principle block diagram of a traffic flow monitoring system for moving objects according to the present invention.
In the traffic flow monitoring system for moving objects according to the present invention, the detection means 101 changes to a moving object in response to the input of a plurality of consecutive frames obtained by photographing the movement of the moving object at predetermined sample intervals. Corresponding images are detected, and the detection results of each frame are subjected to the counting process of the number of moving objects by the counting means 102, and the moving object images included in the detection results of each frame and the preceding frame are identified by the identification means 103. In the moving object traffic flow monitoring system that identifies and uses the identification result for the moving object speed calculation processing by the speed calculating means 104, the speed calculating means 104 includes the sample speed calculating means 125, the weight calculating means 131, and the weighted average. And means 132.

The principle of the moving flow monitoring system for moving objects according to the present invention is as follows.
The sample speed calculation means 125 included in the speed calculation means 104 is a sample of the corresponding moving object at the time corresponding to each frame based on the movement amount on the screen between two consecutive frames of each identified moving object. Calculate the speed. The weight calculation unit 131 included in the speed calculation unit 104 assigns a weight according to the position of the moving object on the screen in each frame based on the correspondence between the position of each pixel in the image of each frame and the position on the shooting target. calculate. The weighted average means 132 provided in the speed calculation means 104 obtains a weighted average of sample speeds corresponding to each of the moving objects based on the weight obtained by the weight calculation means 131, and outputs it as an average speed in the monitoring area.

The operation of the traffic flow monitoring system for moving objects configured in this way is as follows.
According to the first aspect of the present invention, the detection unit 101 detects an image corresponding to the moving object in response to input of a plurality of continuous frames obtained by photographing the moving object moving at predetermined sample intervals. Based on the detection result of each frame, the counting means 102 counts the number of moving objects. In addition, the identification means 103 identifies the moving object image included in the detection result of each frame and the previous frame, and based on this identification result, the speed calculation means 104 calculates the average speed in the section corresponding to the shooting range. Ask. At this time, the weight calculation unit 131 calculates a weight corresponding to the accuracy that can be expected from the sample speed obtained by the sample speed calculation unit 125 for the moving object according to the position occupied by the moving object on the screen. Then, using this weight, the weighted average means 132 weights and averages the sample velocities corresponding to each moving object, thereby taking into account the difference in resolution depending on the position on the screen, and calculating the average speed of the moving object in the monitoring area. It can be calculated accurately.

  According to the present invention, the weight corresponding to the resolution at each position on the screen is weighted and the sample speed is weighted and averaged so that the resolution at each position on the screen is reflected in the average speed in the monitoring area, and the accuracy of the average speed is increased. Therefore, it is possible to provide more accurate information to the user.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
(First embodiment)
FIG. 2 shows a configuration diagram of an embodiment of a traffic flow monitoring system according to the present invention.
2, the traffic flow monitoring system is similar to the traffic flow monitoring system shown in FIG. 7, in accordance with the input of the video captured by the camera 301, the analog-digital conversion processing unit (A / D) 302, the difference The processing unit 303 and the binarization processing unit 305 operate, and the obtained binarized image is used for processing by the detection processing unit 210 and the identification processing unit 220. Further, the processing results of the detection processing unit 210 and the identification processing unit 220 are provided to the processing of the counting processing unit 308 and the speed calculation unit 230.

  In the resolution calculation unit 240 shown in FIG. 2, the table creation unit 241 uses the technique disclosed in Japanese Patent Laid-Open No. 6-47452 to generate a binary image based on the background image held in the still image memory 304. A correspondence table 242 showing the correspondence between the position of each pixel and the position on the road is created, and based on this correspondence table 242, the calculation processing unit 243 indicates the resolution at each position of the binarized image. A resolution table 244 is created and used for processing by the speed calculation unit 230.

In FIG. 2, the detection processing unit 210 includes a detection area setting unit 211, a projection processing unit 212, and a vehicle detection unit 213.
In this detection processing unit 210, the detection region setting unit 211 sets an appropriate detection region in accordance with an instruction from the user, and the projection processing unit 212 performs a projection process described later, and each vehicle detection unit 213. It is the composition which is used for processing.

  In the detection area setting unit 211 shown in FIG. 2, the movement distance calculation unit 214 receives the sample frame interval t and the predicted maximum vehicle speed v from the user, and based on these pieces of information, the vehicle continues. The maximum moving distance d expected between the two sample frames is calculated and sent to the associating unit 215. In response to this, the associating unit 215 refers to the above-described correspondence table 242 to obtain a range of the binarized image having a length corresponding to the above-described movement distance d and starting from the lower side of the screen. May be used as a detection region for detecting a new vehicle.

As described above, the movement distance calculation unit 214 and the association unit 215 operate according to an instruction from the user, so that an appropriate detection region can be obtained from the sampling interval of the binarized image and the predicted maximum speed. Can be set automatically.
Here, since the sampling interval t and the maximum speed v used in the detection area setting unit 211 are objective parameters that do not depend on the subjectivity of the user, it is not necessary to rely on the knowledge and experience of the user. In addition to reducing the burden on the traffic flow, it is possible to accurately detect moving objects such as vehicles by effectively utilizing the processing capacity allocated to the traffic flow monitoring system, and to improve the monitoring accuracy of the traffic flow monitoring system be able to.

Further, in FIG. 2, the projection processing unit 212 performs an area corresponding to one lane divided by a white line (shown with shading in FIG. 3 (a)) as shown in FIG. 3 (a). The pixels in which the difference included in each line is detected (the pixel corresponding to the pixel shown in FIG. 3 (b) with a pixel value of 1) are counted, and the counting result of each line is sent to the vehicle detection unit 213. It is configured to send out.
Here, the counting result of each line obtained in this way indicates the projection result of the binarized image of the region corresponding to each lane described above in the sub-scanning line direction. From this projection result, the original result It is possible to estimate the shape of the vehicle represented by the binarized image.

For example, as shown in FIG. 3 (c), the length of the range in which the counting results equal to or greater than the predetermined threshold Th1 are distributed corresponds to the length of the image representing the vehicle in the binarized image. Features useful for identifying different vehicles.
Therefore, in response to the projection result input for the area corresponding to each lane, the vehicle detection unit 213, for example, sets the range in which the counting results equal to or greater than the threshold Th1 are distributed over a predetermined length. What is necessary is just to detect as a projection result.

As described above, the projection processing unit 212 and the vehicle detection unit 213 can detect the position and characteristics of the vehicle in the current sample frame.
At this time, when the vehicle detection unit 213 detects the projection result of the vehicle from the range indicated by the detection region described above, the number is counted as the number of vehicles newly entering the field of view of the camera 301. In response to this, the counting processing unit 308 may count the number of vehicles that have passed through the field of view of the camera 301 in a certain time, as in the conventional case.

Moreover, this vehicle detection part 213 should just send each position and length of the projection result corresponding to the vehicle detected from the area | region corresponding to each lane to the identification process part 220 as a characteristic of an applicable vehicle.
In the identification processing unit 220 shown in FIG. 2, the vehicle table 221 includes, for example, each vehicle as vehicle information corresponding to each vehicle detected from the image of the previous sample frame for each region corresponding to each lane described above. It is the structure which holds the information showing the characteristic of this.

In addition, the matching processing unit 222 of the identification processing unit 220 uses each of the detection results obtained by the detection processing unit 210 for the new sample frame and the vehicle information held in the vehicle table 221 described above based on conditions described later. It is the structure which collates and identifies the combination which satisfy | fills this condition as the same vehicle.
At this time, the collation processing unit 222 performs collation processing on the condition that the vehicle position obtained as the detection result of the detection processing unit 210 is closest and the vehicle length is equal, and the collation result is obtained as a speed calculation unit. 230 may be sent.

FIG. 4 is a diagram for explaining the identification process.
For example, first, a vehicle that is behind the position of the vehicle indicated by the detection result and whose distance is equal to or smaller than a predetermined threshold Th2 is identified from the vehicle table 221 by a candidate vehicle (reference characters A and B in FIG. 4A). If there are multiple candidate vehicles, select only one candidate vehicle (indicated by symbol a in FIG. 4 (a)) by comparing the detection result with the length of the candidate vehicle. do it.

In addition, the above-described verification processing unit 222 is suitable for the characteristics of the vehicle held in the vehicle table 221 corresponding to the same lane as the vehicle indicated by the detection result according to the input of the detection result of the vehicle. If nothing can be found, similar collation processing may be performed with the vehicle information corresponding to the adjacent lane.
For example, as shown in FIG. 4 (b), the vehicle indicated by the symbol C in FIG. 4 (a) has traveled in two adjacent lanes (FIG. 4) until the next sample frame. In this case, the vehicle indicated by the symbol c is not held in the vehicle table 221 as vehicle information corresponding to the left lane shown in FIG. . In this case, by searching the vehicle table 221 corresponding to the adjacent lane, vehicle information corresponding to the symbol C can be obtained, and the vehicle indicated by the vehicle information and the symbol C in FIG. The vehicle can be identified.

As described above, the collation processing unit 222 of the identification processing unit 220 operates according to the input of the detection result, so that the characteristics of individual vehicles can be identified even when the conventional method considering only the position of the vehicle cannot identify. Considering this, the vehicle can be identified.
This makes it possible to flexibly cope with the driving sequence and sudden speed changes associated with interruptions from adjacent lanes, and to accurately identify vehicles, contributing to improved speed calculation accuracy. Can do.

Further, as described above, each time the identification process for each detection result is completed, the rewrite processing unit 223 displays the vehicle position included in the corresponding vehicle information in the vehicle table 221 in the vehicle in the current sample frame. It is configured to be rewritten at the position.
At this time, the rewrite processing unit 223 directly uses the detection result obtained from the vehicle detection unit 213 as vehicle information regarding the vehicle that has newly entered the field of view of the camera 301 for the new vehicle detected from the detection area described above. What is necessary is just to add to the vehicle table 221 as new vehicle information.

The rewrite processing unit 223 detects vehicle information from the vehicle table 221 when the above-described collation processing unit 222 does not identify the detection result for the current sample frame, and relates to a vehicle that is out of the field of view of the camera 301. What is necessary is just to delete applicable vehicle information as vehicle information.
In this way, the rewrite processing unit 223 rewrites the vehicle table 221 in accordance with the collation result by the collation processing unit 222, so that the information about only the vehicle captured in the field of view of the camera 301 in the current sample frame is stored in the vehicle table. 221 can be used for processing the next sample frame.

The identification result obtained in this way is sent to the speed calculation unit 230, and the speed calculation unit 230 obtains the speed of each vehicle based on the correspondence table 242 obtained by the resolution calculation unit 240 described above. It is done.
In the speed calculation unit 230 shown in FIG. 2, the sample speed calculation unit 231 calculates the sample speed at the position in the current sample frame based on the identification result for each vehicle, as in the conventional case, and the speed holding unit 232. Is sent to the section speed calculation unit 234.

Here, the above-described correspondence table 242 holds the position on the road corresponding to the position of each pixel, and the sample speed calculation unit 231 calculates the sample speed while referring to this correspondence table 242 to obtain the speed. It is configured to send to the holding unit 232.
Accordingly, the speed holding unit 232 may classify the sample speed calculated for each sample frame for each vehicle and hold it together with the position of the vehicle in each sample frame.

In FIG. 2, the weight calculation unit 233 is configured to calculate a weight proportional to the resolution of each pixel on the screen based on the content of the resolution table 244 described above and to use for the processing of the section speed calculation unit 233. ing.
In FIG. 2, the section speed calculation unit 234 obtains a weighted average by adding the weight obtained by the weight calculation unit 233 to each sample speed corresponding to each vehicle held by the speed holding unit 232. Yes.

As described above, the section speed calculation unit 234 and the speed holding unit 232 operate according to the input of the weight from the weight calculation unit 234, thereby realizing the function of the weighted average unit 132 shown in FIG. The result is notified to the center 310 as the section speed of the corresponding vehicle.
At this time, the weight calculation unit 233 reads the position of the corresponding vehicle in each sample frame from the speed holding unit 232 described above, refers to the resolution table 244 based on these positions, and sets the resolution value at each position. An appropriate weight may be calculated based on this and sent to the section speed calculation unit 234.

  For example, the average value of the resolution corresponding to the position in each sample frame and the resolution corresponding to the position in the previous sample frame is simply obtained, and this is multiplied by a predetermined constant α, and the interval speed calculation unit 234 is weighted. Can be sent out. Further, the curve indicating the change in resolution shown in FIG. 8B may be integrated for each section divided by the vehicle position in each sample frame, and the weight in the corresponding section may be obtained based on the integration result.

In this way, the weight calculation unit 233 uses the resolution table 244 obtained by the resolution calculation unit 240 to calculate the weight, thereby realizing the function of the weight calculation unit 131 shown in FIG. The weight corresponding to the resolution corresponding to the position in the obtained image can be used for the average speed calculation process.
In this case, a large weight is applied to the sample speed calculated using the lower image on the screen with high resolution, and conversely, it is added to the sample speed obtained using the image in the low resolution area. The weight is getting smaller.

Accordingly, the resolution at the time of calculating the sample speed can be reflected in the average speed, and the accuracy of calculating the average speed can be improved by utilizing the characteristics of the image information.
As a result, it is possible to notify the center 310 of a more precise average speed for each vehicle, and by analyzing the information at the center 310, accurate information is provided to the user and serviceability is improved. be able to.

The identification process can also be performed using the sample speed obtained as described above.
(Second Embodiment)
FIG. 5 shows a block diagram of a second embodiment of the traffic flow monitoring system according to the present invention.
5, the identification processing unit 220 is configured by adding a position prediction unit 224 to the identification processing unit 220 shown in FIG. 2, and the position prediction unit 224 performs vehicle identification processing for the current sample frame. Prior to this, the position in the binarized image of each vehicle is predicted based on the sample speed of each vehicle received from the speed calculation unit 230 described above, and this prediction result is sent to the collation processing unit 222 via the vehicle table 221. It has a configuration for processing.

The position predicting unit 224 described above receives the sample speed of each vehicle for the previous sample frame from the speed holding unit 232, and on the road of the vehicle in the current sample frame based on the multiplication result of the sample speed and the sample interval t. It is only necessary to calculate the predicted position in the above and further obtain the corresponding position on the screen from the correspondence table 242 described above.
As described above, the position predicting unit 224 operates based on the speed holding unit 232 and the correspondence table 242 to obtain the position on the screen of the vehicle in the current sample frame based on the position of the vehicle in the previous sample frame, It can be used for verification processing.

At this time, the position prediction unit 224 sends the obtained predicted position to the vehicle table 221, and the vehicle table 221 responds accordingly with the predicted position input as part of the vehicle information corresponding to the corresponding vehicle. Then, the matching processing unit 222 may perform identification processing using this predicted position instead of the position in the previous sample frame.
As described above, the vehicle table 221 and the matching processing unit 222 operate in response to the input of the predicted position obtained by the position predicting unit 224, so that information on the speed of the vehicle at the position in the previous sample frame can be obtained. It can be used for identification processing in a frame.

Here, since a vehicle traveling on a road rarely changes its speed suddenly, usually, an image of the vehicle in the current sample frame is detected in the vicinity of the predicted position obtained as described above. It is possible.
Therefore, in this case, when detecting the candidate vehicle, the matching processing unit 222 can narrow down the candidate vehicles using the predetermined threshold Th3 that is significantly smaller than the threshold Th2 described above from the predicted position.

This makes it possible to reliably select only one vehicle by comparing the characteristics (length) of the vehicle, and to identify the vehicle more reliably using the characteristics of the image information. Therefore, it is possible to contribute to improvement in speed calculation accuracy of each vehicle.
Furthermore, if a plurality of features are extracted for each vehicle, the identification accuracy of each vehicle can be further improved.
(Third embodiment)
FIG. 6 shows a configuration diagram of a third embodiment of the traffic flow monitoring system according to the present invention.

  6, the detection processing unit 210 is configured by adding an edge detection unit 216 to the detection processing unit 210 illustrated in FIG. 2, and the edge detection unit 216 receives a binarized image from the binarization processing unit 305. The detected edge-shaped image corresponding to the rear end of the vehicle is detected, and the detected edge length is sent to the identification processing unit 220 together with the detection result by the vehicle detection unit 213 as one of the features of the vehicle. It has a configuration.

  Here, a tail lamp is disposed on the rear side of the vehicle, and an image corresponding to the tail lamp can be easily detected regardless of day and night. Therefore, the edge detection unit 216 adds the tail lamp image to the tail lamp image. Paying attention, an edge-shaped image detection process may be performed. At this time, for example, the edge detection unit 216 detects, as an edge-shaped image, a portion where images corresponding to the tail lamps are arranged at intervals of a predetermined length or more in the horizontal direction, and the length of the edge-shaped image is detected. The information may be sent to the identification processing unit 220 as information indicating the width of the vehicle.

In response to this, the vehicle table 221 of the identification processing unit 220 may hold the width of the vehicle obtained by the edge detection unit 216 and the detection result by the vehicle detection unit 213 in association with each other.
Further, in this case, the collation processing unit 222 of the identification processing unit 220 collates the length and width of the vehicle after searching the candidate vehicle from the vehicle table 221 based on the position in the previous sample frame. The combination of vehicles having the same characteristics may be the same vehicle.

  As a result, even when the vehicle position and length cannot be specified alone, it is possible to select only one candidate vehicle by collating the width of the vehicle, making full use of the characteristics of the image information, The vehicle here can be reliably identified.

It is a principle block diagram of the traffic flow monitoring system of the moving object concerning the present invention. It is a 1st Example block diagram of the traffic flow monitoring system of the moving object concerning this invention. It is a figure explaining a projection process. It is a figure explaining an identification process. It is a 2nd Example block diagram of the traffic flow monitoring system of the moving object concerning this invention. It is a 3rd Example block diagram of the traffic flow monitoring system of the moving object concerning this invention. It is a figure which shows the structural example of the conventional traffic flow monitoring system. It is a figure which shows the correspondence of the position of each pixel on an image, and the position on a road.

Explanation of symbols

101 Detection means 102 Counting means 103 Identification means 104 Speed calculation means 125 Sample speed calculation means 131 Weight calculation means 132 Weighted average means 210, 306 Detection processing section 211 Area setting section 212 Projection processing section 213 Vehicle detection section 214 Distance calculation section 215 Attachment unit 220, 307 Identification processing unit 221 Vehicle table 222 Collation processing unit 223 Rewrite processing unit 224 Position prediction unit 230, 309 Speed calculation unit 231 Sample speed calculation unit 232 Speed holding unit 233 Weight calculation unit 234 Section speed calculation unit 240 Resolution calculation Unit 241 table creation unit 242 correspondence table 243 calculation processing unit 244 resolution table 301 camera 302 analog-digital conversion processing unit (A / D)
303 Difference processing unit 304 Still image memory 305 Binary processing unit 308 Count processing unit 310 Center

Claims (1)

The detection means detects an image corresponding to the moving object in response to an input of images of a plurality of consecutive frames obtained by photographing the moving object moving at predetermined sample intervals, and the detection result of each frame is obtained. The counting means is used for counting the number of the moving objects, and the identification means identifies the images of the moving objects included in the detection results of each frame and the previous frame. In the traffic flow monitoring system for moving objects for speed calculation processing,
Speed calculation means
Sample speed calculating means for calculating the sample speed of the corresponding moving object at the time corresponding to each frame based on the amount of movement on the screen between two consecutive frames of each identified moving object;
A weight calculating means for calculating a weight according to a position on the screen of the moving object in each frame based on a correspondence relationship between the position of each pixel in the image of each frame and the position in the photographing target;
A weighted average means for obtaining a weighted average of sample speeds corresponding to each of the moving objects based on the weights obtained by the weight calculating means, and outputting the average speed in the monitoring area. A traffic flow monitoring system for moving objects.

Images