JPH05282592A - Image processing-type vehicle sensor and method for processing traffic information - Google Patents

Abstract

PURPOSE:To surely confirm the normal connection of a line by a traffic control center. CONSTITUTION:A vehicle on a road is image-picked-up through the use of a video camera 11 so as to overlook the vehicle, the video information such as the travelling position, length and width of the vehicle are continuously processed by an image processing-type vehicle sensor mainbody 12, the persuit and travelling speed, etc., of the vehicle are outputted by a data input/output part 19, the present state of the road is judged and background image data is updated based on the information. Then, a transmission processing part 20 processes communication with the traffic control center. At this time, the traffic control center outputs a test sensing pulse output command so that the image processing-type vehicle sensor mainbody 12 continuously transmits the sensing output of the fixed number of vehicles and fixed pulse width for fixed time to the traffic control center and the traffic control center confirms whether or not abnormality exists in a transmission system based on the information.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is an image processing type vehicle detection device and traffic information installed on a road for measuring and collecting necessary traffic information of vehicle speed, the number of vehicles passing through, small vehicles, large vehicles and the like. Regarding processing method.

[0002]

2. Description of the Related Art Conventionally, necessary traffic information of this kind has been measured,
The vehicle dynamics measurement device that collects images the original image data obtained by capturing the movement of the vehicle on the road and the background image data to extract the traveling vehicle, and extracts the vehicle speed, the number of passing vehicles, the vehicle type (normal vehicle / large vehicle). It outputs measurement data such as car).

[0003]

However, in the image processing type vehicle detection device of the conventional example, necessary traffic information such as the speed of the vehicle, the number of passing vehicles, the type of vehicle (small vehicle / large vehicle), etc. is simply measured and the traffic control is performed. It is being transmitted to the center. For this reason, when the sensor is installed, the determination of normal transmission with the traffic control center is made by transmitting the result of the actual running of the vehicle, and only the traffic control center can accurately determine the normal transmission. There is a problem that you can not.

In the image processing type vehicle detection device of the conventional example, the measurement is mainly performed on the traveling vehicle, and there is a problem that the output of the presence detection cannot be held by detecting the stopped vehicle such as during daytime traffic jam. is there.

In the image processing type vehicle detection device of the conventional example, since a constant threshold value is always used for vehicle type discrimination (large / small), there is a problem that the vehicle type discrimination due to the difference in vehicle brightness cannot be accurately determined.

In the conventional image processing type vehicle detection device, only positive components of the frame difference and the background difference are used in the image processing, but there is a problem that the peripheral portion of the shadow cannot be completely removed.

In the image processing type vehicle detection device of the conventional example, only the plus component of the frame difference and the background difference is used in the image processing, but it is difficult to extract a vehicle with low brightness when the brightness is low such as during twilight. There is.

In the image processing type vehicle detection device of the conventional example, only the positive component of the frame difference and the background difference is used in the image processing. However, the white line on the road may be extracted, and the accurate vehicle width can be determined. There is a problem that it cannot be extracted.

The image processing type vehicle detection device of the conventional example mainly measures the information of the traveling vehicle and has a problem that it cannot detect an illegally parked vehicle or the like outside a predetermined range.

The image processing type vehicle detection device of the conventional example has a problem that when many night vehicles run, it is affected by the road surface reflection of the headlight, and normal background image data cannot be updated.

In the conventional image processing type vehicle detection device, since the vehicle is extracted using only the sample points, it is impossible to keep track of the vehicle in which only the small lamp is turned on at night and the presence detection output when the vehicle is stopped. There is a problem.

The present invention solves such a problem, and provides an image processing type vehicle detection device capable of surely confirming in a traffic control center whether a line is normally connected in accordance with claim 1. The purpose is to do.

It is an object of the present invention to provide an image processing type vehicle detection device which does not hinder traffic control even when a cancellation command is forgotten.

It is an object of the present invention to provide a daytime presence detection signal holding method capable of outputting the presence detection of a stopped vehicle such as during a daytime traffic jam.

It is an object of the present invention to provide a daytime presence detection signal holding method capable of outputting the presence detection of a stopped vehicle such as during a daytime traffic jam.

It is an object of the present invention to provide a daytime vehicle type discriminating method capable of accurately discriminating the vehicle type.

It is an object of the present invention to provide a vehicle extraction method capable of accurately extracting only a vehicle portion.

It is an object of the present invention to provide a method for extracting a vehicle at dusk which can reliably extract a vehicle having low brightness.

It is an object of the present invention to provide a vehicle extraction method capable of removing the influence of a white line drawn on a road.

It is an object of the present invention to provide a vehicle extraction method capable of removing the influence of the white line.

It is an object of the present invention to provide a method for determining whether there is a non-predetermined parking presence, which is capable of adding the function of non-predetermined parking without interfering with conventional functions.

It is an object of the present invention to provide a nighttime background image data updating method capable of stably updating nighttime background image data without being affected by road surface reflection of a headlight.

It is an object of the present invention to provide a night vehicle extraction method capable of reliably extracting headlights.

It is an object of the present invention to provide a night presence detection signal holding method capable of reliably holding the presence detection signal even for a stopped vehicle in which only the small lamp is lit.

It is an object of the present invention to provide a night presence detection signal holding method capable of reliably holding the presence detection signal even for a stopped vehicle in which only the small lamp is lit.

[0026]

In order to achieve this object, an image processing type vehicle sensing device according to claim 1 is a video camera for taking a bird's eye view of the movement of a vehicle on a road, and this video. Image processing the image data of the camera,
It is characterized by comprising an image processing unit for measuring the number of vehicles and the occupancy rate, and a test sensing pulse output means for generating a constant sensing pulse at a constant interval according to a command from the traffic control center.

An image processing type vehicle detection device according to claim 2 is a video camera for taking a bird's eye view of the movement of a vehicle on a road, and image processing of the image data of the video camera to determine the number of vehicles. An image processing unit that measures the occupancy rate,
It is characterized by comprising test sensing pulse output means for generating constant sensing pulses at regular intervals in response to a command from the traffic control center, and automatic release means for the test sensing pulse output means. The daytime presence detection signal holding method according to claim 3 captures an image of a vehicle on the road in a bird's-eye view, processes the image data, measures the number of vehicles and the occupancy rate, and performs sensing measurement. The present invention is characterized in that the distribution of the luminance of the original image data in the area is ternarized using the average luminance value of the background and the presence detection signal is held.

In the daytime presence detection signal holding method according to a fourth aspect, the number of vehicles and the occupancy rate are obtained by performing image processing on the video data of the video camera and the video camera for taking a bird's eye view of the movement of the vehicle on the road. Is measured, and the presence detection signal is held from the difference between the ratio of the most frequent brightness of the background image data in the sensing measurement area and the ratio of the most frequent brightness of the original image data.

A daytime vehicle type discriminating method corresponding to claim 5 is as follows:
Taking a bird's-eye view of the trends of vehicles on the road, image processing this image data, measuring the number of vehicles and occupancy rate, and changing the threshold for vehicle type discrimination according to the brightness of the sensed vehicle. It is a feature.

According to a sixth aspect of the present invention, a vehicle extracting method captures an image of a movement of a vehicle on a road, processes the image data, measures the number of vehicles and the occupancy rate, and calculates the frame difference. In the case of a changed image, it is determined whether there is a change in the frame difference for the previous pixel, whether or not expansion processing is performed behind this pixel, the shadow part is not expanded and the vehicle part It is characterized by expanding only one.

The twilight vehicle extraction method according to claim 7 uses a background difference and a frame difference with expansion processing, in which the movement of the vehicle on the road is photographed so as to give a bird's eye view, and the image data is made effective only with the positive component. It is characterized in that the number of vehicles and the occupancy rate are measured, and the minus component of the background difference is made effective only in the dusk to extract the vehicles with low brightness.

According to the vehicle extraction method of the eighth aspect of the present invention, an image is taken so as to give a bird's eye view of the trend on the road, and the image data is processed by using the background difference and the frame difference with the expansion processing which make only the positive component effective, The number of vehicles and the occupancy rate are measured, and the pixel that has changed by the background difference and the frame difference is also checked for the pixel one pixel before, and only when there is a change in that pixel, the pixel is validated and the noise component It is characterized by reducing.

According to a ninth aspect of the present invention, there is provided a vehicle extraction method for taking a bird's-eye view of the movement of a vehicle on a road, and processing this image data using a background difference and a frame difference with expansion processing that make only positive components valid. In addition to measuring the number of vehicles and occupancy rate, the position of the white line drawn on the road is automatically measured with the background image data created in advance, and image processing is performed excluding the white line part in the measurement area. It is what

The non-predetermined parking existence determining method according to claim 10 captures an image of the movement of the vehicle on the road in a bird's-eye view,
This image data is processed using the background difference that makes only the positive component valid and the frame difference with expansion processing, and the number of vehicles and the occupancy rate are measured, and the lanes are subdivided, and the original image in the subdivided area is processed. It is characterized in that whether or not there is a non-predetermined parking vehicle in the lane is determined based on the most frequent luminance value of the data and the background image data, the ratio thereof, and the average luminance value and the ratio of the pixels that change depending on the frame difference. It is a thing.

According to the method of updating nighttime background image data according to claim 11, a background difference and a frame difference with expansion processing are obtained by taking an image of a vehicle on a road so as to give a bird's eye view, and validating only the plus component of this image data. It is characterized by performing processing by using it to measure the number of vehicles and the occupancy rate, and updating the background image data by making the amount of change of the background image data update in the negative direction larger than the amount of change of the positive component. It is what

According to the twelfth aspect of the present invention, the nighttime vehicle extraction method uses a background difference and a frame difference with expansion processing, in which the movement of the vehicle on the road is taken in a bird's-eye view and the image data is made effective only for the positive component. In addition to processing, measuring the number of vehicles and occupancy rate, image processing is performed on the result of the logical sum of the pixel at the sample point and the pixel next to the sample point only at night to extract the vehicle. is there.

The night presence detection signal holding method according to the thirteenth aspect is such that the movement of the vehicle on the road is photographed in a bird's-eye view, and the background difference and the frame difference with expansion processing for validating only the plus component of this image data are taken. The presence detection signal is processed from the luminance distribution of the result obtained by calculating the logical sum of the pixel of the sample point and the pixel next to the sample point in the sensing output area only at night It is characterized by holding.

According to a fourteenth aspect of the present invention, the night presence detection signal holding method captures an image of a vehicle on the road in a bird's-eye view, and a background difference and a frame difference with expansion processing for validating only the plus component of this image data. The number of vehicles and the occupancy rate are processed, and the distribution of luminance and the background image data of the result of the logical sum of the pixel of the sample point and the pixel next to the sample point in the sensing output area only at night The present invention is characterized in that the presence detection signal is held based on the difference in luminance distribution.

[0039]

With such a configuration, the image processing type vehicle detection device according to claim 1 of the present invention automatically generates the test detection signal in response to a command from the traffic control center. It is possible to reliably confirm the normal connection status at the traffic control center.

The image processing type vehicle detection device according to claim 2 automatically generates a test detection signal in response to a command from the traffic control center, and then automatically operates for a predetermined time without a release command from the traffic control center. Since it is released after the elapse of time, traffic control will not be hindered even if the release command is forgotten.

According to these image processing type vehicle detection devices, confirmation can be performed without arranging an operator in the image processing type vehicle detection device.

In the daytime presence detection signal holding method according to the third aspect, since the histogram of the brightness of the original image data is ternarized using the average brightness value of the background to hold the presence detection signal, the daytime traffic congestion It is possible to output the presence detection of a stopped vehicle such as time.

In the daytime presence detection signal holding method according to the fourth aspect, the presence detection signal is held from the difference between the ratio of the most frequent brightness of the background image data and the most frequent brightness of the original image data. , It is possible to output the presence detection of a stopped vehicle such as during daytime traffic jam.

A daytime vehicle type discriminating method corresponding to claim 5 is as follows:
Since the threshold for vehicle type discrimination is changed according to the brightness of the sensed vehicle, accurate vehicle type discrimination can be performed.

In the vehicle extracting method according to the sixth aspect, since the shadow portion is not expanded from the image data and only the vehicle portion is expanded, only the vehicle portion can be accurately extracted.

In the twilight vehicle extraction method according to the seventh aspect, since the minus component of the background difference is validated during the low twilight luminance time zone, the vehicle with low luminance can be reliably extracted.

In the vehicle extraction method according to the eighth aspect, the pixel which is one pixel before the pixel having a change in the background difference and the frame difference is also confirmed, and only when the pixel one pixel before is changed, the pixel is changed. Is enabled to reduce the noise component, the influence of the white line drawn on the road can be removed.

In the vehicle extraction method according to the ninth aspect, the position of the white line drawn on the road is automatically measured from the background image data created in advance, and the image processing is performed excluding the white line portion in the measurement area. Therefore, the influence of the white line can be removed.

In the non-predetermined parking existence determining method according to the tenth aspect, the lane is subdivided, and the most frequent luminance values of the original image data and the background image data in the subdivided area, their ratio, and the average luminance value. Since the non-predetermined parking in the lane is detected by the ratio of the pixels that change according to the frame difference, the conventional function is not hindered by the conventional hardware alone.
It is possible to add a function for determining the presence of non-prescribed parking.

In the nighttime background image data updating method according to the eleventh aspect, since the amount of change in the negative direction and the amount of change in the positive component of the constant for updating the background image data at night are made larger, Stable background image data can be updated at night without being affected by road reflection.

In the nighttime vehicle extraction method according to the twelfth aspect, the vehicle extraction process is performed only at night using the result of the logical sum of the pixel of the sample point and the pixel adjacent to the sample point. It is possible to reliably extract the light.

In the night presence detection signal holding method according to the thirteenth aspect, since the presence detection signal is held by using the result of the logical sum of the pixel at the sample point and the pixel next to the sample point only at night. The presence detection signal can be reliably held even for a stopped vehicle in which only the small lamp is lit.

According to a fourteenth aspect of the present invention, there is provided a nighttime presence detection signal holding method, wherein only at night, the luminance distribution of the result of obtaining the logical sum of the pixel of the sample point and the pixel adjacent to the sample point in the sensing area and the luminance of the background image data. Since the presence detection signal is held based on the difference from the distribution, it is possible to reliably hold the presence detection signal even for a stopped vehicle in which only the small lamp is lit.

[0054]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of an image processing type vehicle detection device and a traffic information processing method according to the present invention will be described below with reference to the drawings.

First, the first embodiment will be described. The first embodiment corresponds to claim 1. FIG. 1 shows the configuration of the first embodiment. In FIG. 1, reference numeral 11 is a video camera that takes a bird's-eye view of a vehicle on the road, and reference numeral 12 is a video camera that processes image information of the vehicle taken on the road and the speed of the vehicle. This is an image processing type vehicle detection device body that measures and collects necessary traffic information such as the number of vehicles passing through and the type of vehicle (small vehicle / large vehicle). Reference numeral 13 denotes an A / D conversion unit which is built in the image processing type vehicle detection device main body 12 and digitizes the video information, and 14 is built in the image processing type vehicle detection device main body 12 and digitalized by the A / D conversion unit 13. It is an image memory (input image 1) for storing the converted image data. Further, 15 is an image memory (input image 2) which is built in the image processing type vehicle sensing device main body 12 and stores image data digitally converted by the A / D conversion unit 13, and 16 is an image processing type vehicle sensing device. An image memory (input image 3) that is built in the main body 12 and stores image data digitally converted by the A / D converter 13. 1
7 is built in the image processing type vehicle detection device main body 12,
Reference numeral 18 denotes an image memory (input image 4) that stores image data digitally converted by the D conversion unit 13, and 18 is built in the image processing type vehicle sensing device main body 12, and image memories 14 and 1 are provided.
Image data stored in 5 and 16 are processed, the processing result is stored in the image memory 17, the vehicle part is extracted, and necessary traffic information such as vehicle speed, number of passing vehicles, vehicle type (small car / large car), etc. Is an image data processing unit for measuring. Reference numeral 19 denotes a data input / output unit which is built in the image processing type vehicle detection device main body 12 and outputs the traffic information processed by the image data processing unit 18 to a traffic signal controller or the like. Reference numeral 20 denotes a transmission processing unit which is built in the image processing type vehicle detection device main body 12 and performs communication processing with the traffic control center, and transmits various commands to the image processing type vehicle detection device main body 12, The vehicle detection device body 12 performs processing based on the command.

Next, the operation of the configuration according to the first embodiment will be described. In FIG. 1, the video camera 11
Is used to capture an image of the vehicle on the road from a bird's-eye view, and this video information is transmitted to the image processing type vehicle detection device body 12.
In the main body 12 of the image processing type vehicle detection device, the information from here is converted into digital data using the A / D converter 13. Then, the digital data for two screens imaged at a certain fixed interval are stored in the image memories 14 and 15. Further, the image memory 16 stores information (background image data) in a state where no vehicle exists. Then, using the data of these image memories 14, 5 and 6, the image data processing unit 18 uses the background difference method using only the plus component and the frame difference method with the expansion process only on the plus component (by using only the plus component, (Most of which can be removed) is used to write the result of the image processing to the image memory 17, the vehicle is extracted from this image data, and the traveling position, vehicle length, and vehicle width of the vehicle are measured. By continuing this process, the data input / output unit 19 outputs the vehicle tracking, the traveling speed, and the like. This output is connected to the traffic signal controller and can be used for controlling the traffic signal controller, and also determines the current state on the road and updates the background image data based on this information. In addition, the transmission processing unit 20 performs communication processing with the traffic control center. At this time, by outputting a test sensing pulse output command from the traffic control center, the main body 12 of the image processing type vehicle sensing device 12 continuously transmits the sensing output of a certain number and a certain pulse width to the traffic control center at a certain time. At the center,
Based on this information, check whether there is any abnormality in the transmission system.

Next, the test sensing pulse output processing will be described. 2A and 2B show the processing procedure of the test sensing pulse output function. In the communication interrupt of FIG. 2A, information from the center is received (step (in the figure, S
21). Next, it is confirmed whether the received information is a test sensing pulse output start command (step 22). If it is a test sensing pulse output start command,
The test sensing pulse output FLG (flag) is set to 1 (step 23), and the table number of the received test sensing pulse output is stored (step 24). Further, it is confirmed whether or not the received information is a test sensing pulse output stop command (step 25). If the command is the test sensing pulse output stop command, the test sensing pulse output FLG is set to 0 (step 26), and the memory storing the table number of the received test sensing pulse output is cleared (step 27). Also, it is confirmed whether the received information is another command (step 2 pulse). If it is another command, the process according to the command is performed (step 29).

Next, in FIG. 2B, based on the result of the communication interrupt process, in the normal process, it is first checked whether the test sensing pulse output FLG is 1 (switch 210). If FLG is not 1, normal sensing processing is performed (step 211). If FLG is 1, first, the memory that stores the table number of the test sensing pulse output is read (step 212). Then, the test sensing pulse corresponding to the table number is output (step 213).

As described above, in the first embodiment, the test control signal is automatically generated according to the command from the traffic control center, and the normal connection state of the line can be surely confirmed at the traffic control center. Further, the confirmation can be performed without arranging a worker in the image processing type vehicle detection device.

Next, the test sensing pulse output stop processing of the second embodiment will be described. This second embodiment corresponds to claim 2.

FIGS. 3A, 3B and 3C show the procedure of the test sensing pulse output stop processing. In the communication interruption shown in FIG. 3A, the information from the center is received (step 31). Next, it is confirmed whether or not the received information is a test sensing pulse output start command (step 32). If the command is a test sensing pulse output start command, the test sensing pulse output FLG is set to 1 (step 33) and the table number of the received test sensing pulse output is stored (step 34). Further, it is confirmed whether or not the received information is a test sensing pulse output stop command (step 35). If the command is the test sensing pulse output stop command, the test sensing pulse output FLG is set to 0 (step 3).
Along with 6), the memory that stores the table number of the received test sensing pulse output is cleared (step 3
7). Further, it is confirmed whether the received information is another command (step 38). If it is another command, the process according to the command is performed (step 39).

Next, in the timer interrupt processing shown in FIG. 3B, it is first confirmed whether or not the test sensing pulse output FLG is 1 (step 310). FLG is 1
If not, the test sensing pulse output counter is cleared (step 311), and if FLG is 1, the test sensing pulse output counter is incremented (step 31).
2). Then, it is confirmed whether or not this counter is larger than the set value (step 313).
The test sensing pulse output FLG is forcibly set to 0 (step 314), and the memory storing the table number of the received test sensing pulse output is cleared (step 315).

From the results of these interrupt processes, FIG.
In the normal processing shown in (c), it is first checked whether the test sensing pulse output FLG is 1 (step 3).
16). If FLG is not 1, normal sensing processing is performed (step 317). If FLG is 1, first, the memory that stores the table number of the test sensing pulse output is read (step 318). Then, the test sensing pulse corresponding to the table number is output (step 31).
9).

As described above, in the second embodiment, after a test sensing signal is automatically generated in response to a command from the traffic control center, if a predetermined time elapses even if there is no cancellation command from the traffic control center. To release. For this reason, traffic control will not be hindered even if the cancellation command is forgotten.
Further, the confirmation can be performed without arranging a worker in the image processing type vehicle detection device.

Next, the presence sensing output in the third embodiment will be described. This third embodiment corresponds to claim 3.

FIGS. 4 (a), (b), (c) and (d) show areas for measuring the presence sensing output. In FIG. 4A, the last 4 to 6 m of the normal measurement area is set as the presence detection measurement area, and the original image data of this area, the average luminance of the background image data, the most frequent value of luminance, etc. are measured. Measure the presence detection output of each lane in the daytime.

Normally, as shown in FIG. 4 (b), the luminance distribution of the original image data in the sensing and focusing area is output at a substantially constant luminance. However, as shown in FIG. 4C, when the vehicle exists in the sensing / measurement area, the luminance distribution is distorted. By using this condition, as shown in FIG. 4D, the luminance distribution of the original image data is ternarized with the average luminance value of the background image data as a boundary, and the number of pixels determined to be −1 and −1. The presence / absence of a vehicle is determined based on the ratio of the number of pixels determined to be 0, and the presence detection output is held.

Further, the daytime presence detection signal holding process will be described. FIG. 5 shows a procedure of a daytime presence detection signal holding process. In FIG. 5, first, a processing screen is created using the background subtraction method using only the plus component and the frame subtraction method with expansion processing only on the plus component (step 51). Next, the traveling vehicle is extracted from this processing screen (step 52). Then, the extracted vehicle is processed to perform a vehicle tracking process (step 53). Further, it is determined whether or not the tracking vehicle has entered the sensing / measurement area (step 54).
When it is determined that the tracking vehicle has entered, the presence detection output is turned on (step 55). Next, the average luminance of the background image data is calculated (step 56), and the current day / night state is determined (step 57). In the case of night, a normal presence detection holding process is performed (step 58). In the daytime, first, the luminance distribution (histogram) of the original image data in the sensing measurement area is measured (step 59). Next, based on the average luminance value of the background image data, the luminance distribution of the original image data in the sensing measurement area is ternarized (step 510), and the number of pixels near the average luminance of the background image data is set to another number of pixels. Is determined (step 51).
1). If it does not exceed, presence detection output is turned off.
(Step 512).

As described above, according to the third embodiment, the histogram of the luminance of the original image data in the sensing / measurement area can be ternarized by using the average luminance value of the background to hold the presence sensing signal. , It can output the presence detection of a stopped vehicle, such as during daytime traffic jams.

Next, the daytime presence detection signal holding processing in the fourth embodiment will be described. The fourth embodiment corresponds to claim 4.

FIGS. 6A, 6B and 6C show measurement areas for holding the presence detection signal in the daytime. Figure 6 (a)
, The last 4 to 6 m of the normal measurement area is set as the presence detection measurement area, and the average luminance of the original image data, the background image data, the most frequent value of the luminance, etc. of this area are measured to determine the daytime of each lane. Measure the presence sensing output.

Normally, as shown in FIG. 6B, the luminance distribution of the original image data in the sensing / measurement area is concentrated at a substantially constant luminance. However, as shown in FIG. 6C, when the vehicle exists in the sensing measurement area, the luminance distribution is distorted. By using this condition, the presence / absence of the vehicle can be determined by comparing the ratio of the most frequent brightness value of the original image data with the ratio of the most frequent brightness value of the background image data, and the presence sensing output can be held.

Further, the daytime presence detection signal holding process will be described. FIG. 7 shows the procedure for holding the presence detection signal. In FIG. 7, first, a processed screen is created using the background subtraction method using only the plus component and the frame subtraction method with expansion processing only on the plus component (step 71).
Next, the traveling vehicle is extracted from this processing screen (step 72). Further, the extracted vehicle is processed to perform vehicle tracking processing (step 73). Further, it is determined whether or not the tracking vehicle has entered the sensing / measurement area (step 74).
When it is determined that the tracking vehicle has entered, the presence detection output is turned on (step 75). Next, the average luminance of the background image data is calculated (step 76), and the current day / night state is determined (step 77). In the case of nighttime, normal presence detection holding processing is performed (step 78). In the daytime, first, the luminance distribution (histogram) of the original image data in the sensing measurement area is measured (step 79). next,
The luminance distribution (histogram) of the background image data in the sensing measurement area is measured (step 710). Next, the ratios of the most frequent luminance values of the background image data and the original image data in the sensing measurement area are compared (step 711). If the ratio is equal to or more than the threshold value, the presence sensing output is turned off (step 712). ).

As described above, according to the fourth embodiment, the presence detection signal can be held from the difference between the ratio of the most frequent brightness of the background image data and the ratio of the most frequent brightness of the original image data. It can output the presence detection of a stopped vehicle.

Next, a fifth embodiment will be described. The fifth embodiment corresponds to claim 5.

FIG. 8 shows a processing procedure for daytime vehicle type discrimination. In FIG. 8, first, the inside of the vehicle tracking area is subjected to image processing of frame difference with expansion processing + background difference to create a processing screen (step 81). Next, the vehicle portion is extracted from the processing screen (step 82). Next, the extraction result on the processing screen is analyzed to extract the position of the front edge portion of the vehicle (step 83). Then, the leading edge data is compared with the leading edge portion and speed data up to the previous time, and the vehicle tracking processing is performed (step 84). For the vehicle that has been tracked, the vehicle speed, the pulse width of the presence detection signal, etc. are calculated (step 85). Next, the vehicle type is determined. First, the average luminance of the background image data is obtained (step 86), and the current day / night state is determined (step 87). In the case of nighttime, a vehicle type discrimination process is performed using the normal width between headlights (step 88). In the daytime, first, the vehicle length and the vehicle width are measured from the vehicle extraction portion on the processing screen (step 8).
9). Then, the position of the vehicle extraction portion on the processing screen is measured (step 810). Next, the average brightness of the vehicle extracted portion in the original image data is calculated (step 811). Then, it is determined whether this average brightness is brighter than a threshold value (step 812). The threshold value (vehicle length, vehicle width) for vehicle type discrimination is set to two levels according to the brightness, and vehicle type discrimination is performed (steps 313 and 314). These results are transmitted to the traffic control center (step 815). Also, the background image data is updated according to the condition on the road (step 81).
6).

As described above, according to the fifth embodiment, it is possible to accurately determine the vehicle type by changing the vehicle type determination threshold value according to the brightness of the sensed vehicle.

Next, a sixth embodiment will be described. The sixth embodiment corresponds to claim 6.

9 (a) (b) (c) (d) (e)
(F) is a conceptual diagram using only the plus component of the frame difference with expansion processing + the background difference. Figure 9
As shown in (a), (b) and (c), first, the image data is t-Δt [sec], t [sec] image data,
There are three types of background image data. First, t-Δt
The difference between the image data of [sec] and t [sec] is calculated.
This is called a frame difference as shown in FIG. 9D, and if only the positive component is made effective, the front edge portion of the traveling vehicle and the rear portion of the shadow beside the vehicle are extracted.

Since the moving object can be surely extracted from the frame difference, the expansion processing is performed rearward (lane portion in FIG. 9D) to make it more reliable. Further, the difference between the image data of t [sec] and the background image data is calculated. Figure 9
As shown in (e), it is called a background difference, and when only the positive component is enabled, the shadow portion is not extracted and only the bright portion of the vehicle is extracted. Then, as shown in FIG.
By extracting the logical sum of the frame difference and the background difference and creating a processing screen, the vehicle can be reliably extracted. For bright white vehicles, extraction is performed from both the frame difference and background difference.
In the case of a dark black vehicle, the brightness is basically higher even if the same vehicle is located closer to the front of the image, so that the vehicle can be extracted by the frame difference. Further, although the stopped vehicle cannot be extracted by the frame difference, it is extracted by the background difference.

However, as shown in FIG. 9D, there is a problem that the shadow portion is also expanded. Therefore, only the vehicle portion is inflated, and the shadow portion is not inflated, by setting the condition for inflating. This will be described with reference to FIGS. 10 (a), (b) and (c).

As shown in FIG. 10A, in the normal expansion process, when a pixel changes due to the frame difference, the expansion process is performed. In this case, as shown in FIG. 10B, when a pixel changes due to the frame difference, it is confirmed whether the pixel of the previous scanning line changes due to the frame difference, and the expansion process is performed only when the change occurs. To do so.

As shown in FIG. 9, most of the shadow is removed, but part of the latter half remains. Here, as shown in FIG. 10C, by performing the conditional expansion processing, the expansion of the shadow portion is surely reduced.

Next, the processing procedure of the vehicle extraction method will be described. FIG. 11 shows a processing procedure of the vehicle extraction method. In FIG. 11, first, image data t [sec] and t-Δ
The difference (frame difference) of t [sec] is calculated, and only the plus component is validated (step 111). Subsequently, the difference between the image data t [sec] and the background image data (background difference) is performed, and only the plus component is validated (step 1).
12). Then, first, the frame difference is compared with the data threshold value α1 for each pixel (step 114). If the threshold value is equal to or greater than α1, it is determined whether the pixels in the same row of the 1st and 2nd scanning lines are changed by the frame difference (step 116,
117). If there is a change in both, expansion processing is performed in which 1 is written in the same position on the processed screen and the pixels in the same column before 3, 4, and 5 scanning lines (step 118). If there is no change,
Write 1 only to the same position on the machining screen (step 11
10). If the pixel difference data of the frame difference is the threshold value α
In the case of 1 or less, the data for each pixel of the background difference is set to the threshold α2
And (step 119). If the threshold is 2 or more, 1 is written at the same position on the processing screen (step 1110). In other cases, 0 is written at the same position on the processing screen (step 1111). This is processed for all pixels to create a processed screen.

As described above, according to the sixth embodiment, only the vehicle portion can be accurately extracted by expanding only the vehicle portion without expanding the shadow portion from the image data.

Next, a seventh embodiment will be described. The seventh embodiment corresponds to claim 7. FIG. 12 shows a processing procedure of a vehicle extraction method at dusk. In FIG. 12, first, image data t [sec] and t-Δt [se]
c] (frame difference) is performed, and only the positive component is validated (step 121). Then, the image data t
The difference between [sec] and the background image data (background difference) is calculated (step 122). Next, the average brightness of the background image data is obtained (step 123). By comparing the average luminance of the background image data, it is determined whether the current state is daytime, dusk, or night (step 124), and in the case of daytime or nighttime, a normal process for validating only the plus component is performed. Perform (step 125). In the case of twilight, first, the data of each pixel of the frame difference is compared with the threshold value α1 (step 128). Then, when the value is equal to or larger than the threshold value α1, it is determined whether or not the pixels in the same column of the 1st and 2nd scanning lines are the change image by the frame difference (steps 129 and 1210). If both are changed, expansion processing for writing 1 to the same position on the processed screen and pixels on the same row before 3, 4, and 5 scanning lines is performed (step 1211). Only 1 is written (step 1211). If the pixel difference data of the frame difference is less than or equal to the threshold value α1, the absolute value of the data of each pixel of the background difference is converted into the threshold value α2.
Is compared (step 1212). Then, the threshold α2
In the above case, 1 is written at the same position on the processing screen (step 1213). In other cases, 0 is written in the same position on the processing screen (step 1214). This is processed for all pixels to create a processed screen.

As described above, according to the seventh embodiment, the minus component of the background difference is also made effective in the time zone of low brightness such as during twilight, so that even the vehicle of low brightness can be reliably extracted. ..

Next, an eighth embodiment will be described. The eighth embodiment corresponds to claim 8.

FIG. 13 shows the processing procedure of the vehicle extraction method. In FIG. 13, first, image data t [sec]
And t-Δt [sec] (frame difference),
Only the plus component is validated (step 131). Then, the difference between the image data t [sec] and the background image data (background difference) is performed, and only the plus component is validated (step 132). Then, first, the data of each pixel of the frame difference is compared with the threshold value α1 (step 135).
When the threshold value is equal to or larger than α1, the pixel one pixel before is 2
It is determined whether it is 0h or 80h (step 136), and if it is 20h or 80h, 80h is written at the same position on the machining screen and expansion processing is performed (step 137), and 20h.
If it is not 80h, 20h is written at the same position on the processing screen (step 138). If the data for each pixel of the frame difference is less than or equal to the threshold value α1, the data for each pixel of the background difference is compared with the threshold value α2 (step 13).
9). If it is equal to or larger than the threshold value α, it is determined whether the pixel one pixel before is 20h or 80h (step 131).
In the case of 0) and 20h, 80h is written at the same position on the working screen (step 1311), and 20h is written at the same position on the working screen (step 138). In other cases, 0 is written at the same position on the processed screen (step 1312), and 0 is written at the position of the pixel one pixel before the processed screen (step 1313). This is processed for all pixels to create a processed screen.

As described above, according to the eighth embodiment, with respect to the pixel having a change in the background difference and the frame difference, the pixel one pixel before is also confirmed, and only when the pixel one pixel before is changed. , The pixel is validated to reduce the noise component, so that the influence of the white line drawn on the road can be removed.

Next, a ninth embodiment will be described. The ninth embodiment corresponds to claim 9.

14A and 14B show the processing procedure of the vehicle extraction method. In the initial processing of FIG. 14A, necessary constants such as the measurement area, the position of the lane, and the focal length of the CCD camera are input using a mouse, a keyboard or the like (step 141). Next, a certain fixed screen original image data is imaged, the brightness distribution and the brightness average for each pixel are obtained, and the initial background image data is created (step 142). Further, the initial background image data is subjected to a differential process in the horizontal direction to create a processing screen for extracting a white line portion. (Step 143). In this case, the changed pixels and the pixels in the vicinity of the boundary between the measurement area and the lane are not used for the processing such as the vehicle extraction processing (step 144). Then, a normal measurement process is performed. Image processing of frame difference with expansion processing + background difference of pixels excluding the pixel portion of the white line portion in the measurement area is performed to create a processing screen (step 145). Next, the vehicle portion is extracted from the processing screen (step 146). Also, the extraction result on the processing screen is analyzed to extract the position of the front edge portion of the vehicle (step 14).
7). And this leading edge data and the leading edge part up to the last time,
The speed data are compared and a vehicle tracking process is performed (step 148). Then, for the vehicle whose tracking has been completed, the presence detection signal, the vehicle speed, and the vehicle type are calculated (step 14).
9). Then, these results are transmitted to the traffic control center (step 1410). Also, the background image data is updated according to the state on the road (1411).

As described above, according to the ninth embodiment, the position of the white line drawn on the road is automatically measured from the background image data created in advance, and the image processing is performed excluding the white line portion in the measurement area. As a result, the influence of the white line can be removed.

Next, a tenth embodiment will be described. In addition,
This tenth embodiment corresponds to claim 10.

FIG. 15 is a diagram showing an area in which the measurement of the illegal parking existence determination is performed. In FIG. 15, by dividing the area into lanes near the sidewalk every 4 to 5 m in the normal measurement area and measuring the original image data, the average luminance of the background image data, and the most frequent luminance data by measuring each area. Measure the presence of illegal parking.

Next, a method for determining the presence of illegal parking will be described. FIG. 16 shows a processing procedure of the illegal parking existence determining method. In FIG. 16, first, a normal vehicle tracking process is performed to measure the vehicle (step 161). Next, the lane of the original image data that touches the sidewalk is subdivided to measure the average luminance and the most frequent luminance data for each illegal parking measurement area (step 162). Further, with respect to the background image data, the average luminance and the most frequent luminance data are similarly measured for each illegal parking measurement area (step 163). Then, it is determined whether the front edge of the tracked vehicle exists in each area (step 165). If there is a tracking vehicle in the illegal parking measurement area, it is determined that no stopped vehicle exists in the area (step 166). If there is no tracking vehicle, the average brightness in the area where the presence detection output of the original image data and the background image data is measured is compared (step 167). And if there is a brightness difference,
It is determined that there is a stopped vehicle in the area (step 168). If there is no brightness difference, the most frequent brightness data in the measurement area of the original image data and the background image data is compared (step 169). If there is a brightness difference, it is determined that there is a stopped vehicle in the area (step 168). If it does not exist, it is determined that there is no stopped vehicle (step 166). This information is time-sequentially measured for each area (step 1611), and it is determined whether or not a stopped vehicle exists in the same area for a predetermined time or longer (step 1612). If it exists, the fact that an illegally parked vehicle exists is transmitted to the traffic control center (step 1613). Also, the background image data is updated according to the state on the road (1613).

As described above, according to the tenth embodiment, the lanes are subdivided, and the most frequent luminance values and their ratios of the original image data and the background image data in the subdivided area, the average luminance value and the frame difference. By detecting the illegal parking of the lane in the ratio of the pixels that have changed, it is possible to add the presence determination function of the illegal parking without hindering the conventional function only by the conventional hardware.

Next, an eleventh embodiment will be described. In addition,
This eleventh embodiment corresponds to claim 11.

FIG. 17 shows the processing procedure of the nighttime background image data updating method. In FIG. 17, first, the inside of the vehicle tracking area is subjected to image processing of frame difference with expansion processing + background difference to create a processing screen (step 17).
1). Next, the vehicle portion is extracted from the processing screen (step 172). Furthermore, the extraction result on the processing screen is analyzed to extract the position of the front edge portion of the vehicle (step 17).
3). And this leading edge data and the leading edge part up to the last time,
The speed data is compared and a vehicle tracking process is performed (step 174). For a vehicle that has been tracked, a presence detection signal and vehicle speed are calculated (step 175). And
These results are transmitted to the traffic control center (step 176). Also, the background image data is updated according to the condition on the road. The average brightness of the background image data is calculated (step 177), and the current day / night state is judged (step 1).
78). In the daytime, normal background image data is updated (step 179). In the case of nighttime, the original image data and the background image data are compared in units of brightness (step 171).
0), if the original image data is higher, the pixel brightness of the background image data is incremented by 1 (step 171).
2) If the background image data is higher, the pixel brightness of the background image data is decremented by 2 (step 171).
3).

As described above, according to the eleventh embodiment, by making the amount of change in the negative direction of the constant for updating the background image data at night larger than the amount of change in the positive component, the influence of the road surface reflection of the headlight is affected. It is possible to update the background image data stably at night without receiving the data.

Next, a twelfth embodiment will be described. In addition,
This twelfth embodiment corresponds to claim 12.

FIG. 18 shows a processing procedure of the night vehicle extraction method. In FIG. 18, first, image data t [se
The difference (frame difference) between c] and t-Δt [sec] is calculated, and only the positive component is validated (step 181).
Then, the difference between the image data t [sec] and the background image data (background difference) is performed. The average brightness of the background image data is calculated (step 182), and the current day / night state is determined (step 183). In the case of night, a normal difference is made for each pixel, and a process for validating only the plus component is performed (step 184). In the case of night, a logical sum of the pixel serving as the sample point of the original image data and the pixel one pixel before is calculated (step 186), the result is compared with the background image data, and only the plus component is validated. Is performed (step 187). Then, first, the data for each pixel is compared with a threshold for the processing result of the frame difference. Then, when it is equal to or larger than the threshold value, 1 is written in the same position on the processing screen.
Then, it is determined whether or not the pixels in the same column of the 1st and 2nd scanning lines change due to the frame difference, and if there is a change in both, write 1 to the pixels in the same column before the 3rd, 4th and 5th scanning lines. Processing is performed (step 1810). Further, the data for each pixel is compared with a threshold value based on the background difference processing result. Then, when it is equal to or larger than the threshold value, 1 is written in the same position on the processing screen. In other cases, 0 is written in the same position on the processing screen (step 1811). This processing is performed to create a processing screen.

As described above, according to the twelfth embodiment, the vehicle extraction processing is performed by using the result of the logical sum of the pixel at the sample point and the pixel adjacent to the sample point only at night, and thus the headlight Extraction can be reliably performed.

Next, a thirteenth embodiment will be described. In addition,
This thirteenth embodiment corresponds to claim 13.

FIG. 19 shows a processing procedure of the night presence detection signal holding method. In FIG. 19, first, image processing of the frame difference + background difference with expansion processing is performed in the vehicle tracking area to create a processing screen (step 191).
Next, the vehicle portion is extracted from the processing screen (step 1
92). Next, the extraction result on the processing screen is analyzed to extract the position of the front edge portion of the vehicle (step 193). Then, the leading edge data is compared with the leading edge portion and the speed data up to the previous time, and the vehicle tracking processing is performed (step 194).
Next, it is determined whether or not the tracking vehicle has entered the sensing / measurement area (step 195). If it is determined that the tracking vehicle has entered, the presence detection output is turned on (step 196). Further, the presence detection signal length, vehicle speed, vehicle width, etc. are calculated (step 197). Next, presence detection signal holding processing is performed. The average brightness of the background image data is calculated (step 198), and the current day / night state is judged (step 199). In the daytime, a normal presence detection holding process is performed (step 1910). In the case of night, the logical sum of the pixel serving as the sample point of the original image data in the sensing measurement area and the pixel one pixel before is obtained (step 191).
1), luminance distribution (histogram) using this data
Is measured (step 1912). Then, it is determined whether or not there are more than the threshold number of pixels in this luminance distribution (step 1913). If there is not more than the threshold value, the presence detection output is turned off (step 1914). Also, the background image data is updated according to the condition on the road (step 1915).

As described above, according to the thirteenth embodiment, the presence sensing signal is held by using the result of the logical sum of the pixel at the sample point and the pixel next to the sample point in the sensing measurement area only at night. As a result, the presence detection signal can be reliably held even for a stopped vehicle in which only the small lamp is lit.

Next, a fourteenth embodiment will be described. In addition,
This fourteenth embodiment corresponds to claim 14.

FIG. 20 shows a processing procedure of the night presence detection signal holding method. In FIG. 20, first, image processing of frame difference + background difference with expansion processing is performed in the vehicle tracking area to create a processing screen (step 201).
Next, the vehicle portion is extracted from the processing screen (step 2).
02). Subsequently, the extraction result on the processing screen is analyzed to extract the position of the front edge portion of the vehicle (step 203). Then, the leading edge data is compared with the leading edge portion and the speed data up to the previous time, and the vehicle tracking processing is performed (step 204).
Next, it is determined whether or not the tracking vehicle has entered the sensing / measurement area (step 205). When it is determined that the tracking vehicle has entered, the presence detection vehicle is turned on (step 206). And presence detection signal length, vehicle speed,
The vehicle width and the like are calculated (step 207). Further, presence detection signal holding processing is performed. The average luminance of the background image data is calculated (step 208), and the current day / night state is determined (step 209). In the case of daytime, a normal presence detection holding process is performed (step 2010). In the case of nighttime, the logical sum of the pixel which is the sample point of the original image data in the sensing measurement area and the pixel one pixel before is obtained (step 2011), and the distribution of luminance (histogram) is obtained using this data. Measure (step 2012). In addition, the luminance distribution (histogram) in the sensing measurement area of the background image data is measured (step 2013). Subsequently, the difference between the respective brightness distributions is performed (step 2014), it is determined whether or not there are more than a threshold number of pixels having a brightness equal to or higher than the threshold value in the difference result (step 2015). The presence detection output is turned off (step 2016). Further, the background image data is updated according to the condition on the road (step 2017).

As described above, according to the fourteenth embodiment, the luminance distribution of the result of the logical sum of the pixel at the sample point and the pixel next to the sample point in the sensing measurement area only at night and the luminance of the background image data are obtained. By holding the presence detection signal from the difference from the distribution, it is possible to reliably hold the presence detection signal even for a stopped vehicle in which only the small lamp is lit.

[0110]

As is apparent from the above description, the image processing type vehicle detection device according to the first aspect of the present invention automatically generates the test detection signal in response to a command from the traffic control center. Therefore, there is an effect that the normal state of the line can be surely confirmed at the traffic control center.

The image processing type vehicle detection device according to claim 2 automatically generates a test detection signal in response to a command from the traffic control center, and then automatically operates for a predetermined time without a release command from the traffic control center. Since the cancellation is made after the lapse of time, there is an effect that traffic control is not hindered even if the cancellation command is forgotten.

According to these image processing type vehicle detection devices, confirmation can be performed without arranging a worker in the image processing type vehicle detection device.

In the daytime presence detection signal holding method according to the third aspect, since the presence histogram detection signal is held by ternarizing the histogram of the brightness of the original image data by using the average background brightness value, the daytime traffic jam occurs. This has the effect that the presence detection of a stopped vehicle can be output at times.

In the daytime presence detection signal holding method according to the fourth aspect, the presence detection signal is held from the difference between the ratio of the most frequent brightness of the background image data and the most frequent brightness of the original image data. In addition, it is possible to output the presence detection of a stopped vehicle such as during daytime traffic congestion.

A daytime vehicle type discriminating method according to claim 5 is as follows:
Since the threshold value for vehicle type discrimination is changed according to the brightness of the sensed vehicle, there is an effect that accurate vehicle type discrimination can be performed.

In the vehicle extracting method according to the sixth aspect, since the shadow portion is not expanded from the image data and only the vehicle portion is expanded, only the vehicle portion can be accurately extracted.

In the twilight vehicle extraction method according to the seventh aspect, since the minus component of the background difference is made effective during the low twilight luminance time zone, it is possible to reliably extract the vehicle with low luminance. ..

In the vehicle extraction method according to the eighth aspect, the pixel which is one pixel before the pixel having a change in the background difference and the frame difference is also confirmed, and the pixel is changed only when the pixel one pixel before is changed. Is enabled to reduce the noise component, the effect of removing the white line drawn on the road can be removed.

In the vehicle extraction method according to claim 9, the position of the white line drawn on the road is automatically measured from the background image data created in advance, and the image processing is performed excluding the white line portion in the measurement area. Therefore, there is an effect that the influence of the white line can be removed. In the non-predetermined parking existence determining method according to claim 10, the lane is subdivided, and the most frequent luminance values of the original image data and the background image data in the subdivided area, their ratio, the average luminance value, and the frame difference. Since the non-scheduled parking is detected in the lane in the ratio of the pixels that change in, the effect that the presence determination function of the non-scheduled parking can be added without interfering with the conventional function only by the conventional hardware is provided. ..

In the nighttime background image data updating method according to the eleventh aspect, since the amount of change in the negative direction and the amount of change in the positive component of the constant for updating the background image data at night are made larger, This has an effect that stable background image data can be updated at night without being affected by road reflection.

In the nighttime vehicle extraction method according to the twelfth aspect, the vehicle extraction process is performed only by using the result of the logical sum of the pixel of the sample point and the pixel adjacent to the sample point only at night. This has the effect that the headlights can be reliably extracted.

In the night presence detection signal holding method according to the thirteenth aspect, since the presence detection signal is held by using the result of the logical sum of the pixel at the sample point and the pixel adjacent to the sample point only at night, This has an effect that the presence detection signal can be surely held even for a stopped vehicle in which only the small lamp is lit.

According to a fourteenth aspect of the present invention, there is provided a nighttime presence detection signal holding method, wherein the luminance distribution of the result of obtaining the logical sum of the pixel of the sample point and the pixel adjacent to the sample point in the sensing area only at night and the luminance of the background image data Since the presence detection signal is held based on the difference from the distribution, it is possible to reliably hold the presence detection signal even for a stopped vehicle in which only the small lamp is lit.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an image processing type vehicle detection device (first embodiment) of the present invention.

2A and 2B are flowcharts showing a processing procedure of a test sensing pulse output function in the first embodiment.

3 (a), (b) and (c) are flow charts showing a procedure of a test sensing pulse output stopping process of an image processing type vehicle sensing device (second embodiment) according to claim 2 of the present invention.

4 (a), (b), (c) and (d) show areas for measuring presence detection output in a daytime presence detection signal holding method (third embodiment) corresponding to claim 3 of the present invention. Illustration

FIG. 5 is a flowchart showing a procedure of a daytime presence detection signal holding process in the third embodiment.

6 (a), (b) and (c) show measurement areas for holding the presence detection signal in the daytime in the method for holding the presence detection signal in the daytime (fourth embodiment) according to claim 4 of the present invention. Illustration

FIG. 7 is a flow chart showing the procedure of holding presence detection signal in the fourth embodiment.

FIG. 8 is a flowchart showing a processing procedure of a daytime vehicle type identification method (fifth embodiment) corresponding to claim 5 of the present invention.

9 (a), (b), (c), (d), (e), and (f) are frame differences with expansion processing of the vehicle extraction method (sixth embodiment) corresponding to claim 6 of the present invention. + Conceptual diagram using only plus component of background difference

10 (a), (b), and (c) are explanatory views when only the vehicle portion in the sixth embodiment is inflated and the shadow portion is not inflated.

FIG. 11 is a flowchart showing a processing procedure of a vehicle extraction method in the sixth embodiment.

FIG. 12 is a flowchart showing a processing procedure of a twilight vehicle extraction method (seventh embodiment) corresponding to claim 7 of the present invention.

FIG. 13 is a flowchart showing a vehicle extraction method (eighth embodiment) processing procedure according to claim 8 of the present invention.

14 (a) and 14 (b) are flowcharts showing a processing procedure of a vehicle extraction method (ninth embodiment) corresponding to claim 9 of the present invention.

FIG. 15 is a diagram showing an area for measuring an illegal parking presence determination in the non-predetermined parking presence determination method (tenth embodiment) according to claim 10 of the present invention.

FIG. 16 is a flowchart showing a processing procedure of an illegal parking existence determining method according to the tenth embodiment.

FIG. 17 is a flowchart showing a processing procedure of a nighttime background image data updating method in a nighttime background image data updating method (eleventh embodiment) according to claim 11 of the present invention.

FIG. 18 is a flowchart showing a processing procedure of a night vehicle extraction method in a night vehicle extraction method (twelfth embodiment) according to claim 12 of the present invention.

FIG. 19 is a flowchart showing a processing procedure of a night presence detection signal holding method in a night presence detection signal holding method (thirteenth embodiment) corresponding to claim 13 of the present invention.

FIG. 20 is a flowchart showing the processing procedure of the night presence detection signal holding method in the night presence detection signal holding method (14th embodiment) according to claim 14 of the present invention.

[Explanation of symbols]

 11 Video Camera 12 Image Processing Vehicle Sensing Device Main Body 13 A / D Converter 14, 15, 16, 17 Image Memory 18 Image Data Processor 19 Data Input / Output Unit 20 Transmission Processor

Claims (14)

[Claims] 1. A video camera for taking a bird's-eye view of the movement of vehicles on the road, an image processing unit for processing the image data of the video camera to measure the number of vehicles and the occupancy rate, and a traffic control center. An image processing type vehicle detection device, comprising: a test detection pulse output means for generating a constant detection pulse at a constant interval in response to a command from the image processing vehicle. 2. A video camera for taking a bird's-eye view of the movement of vehicles on the road, an image processing unit for processing the image data of the video camera to measure the number of vehicles and the occupancy rate, and traffic control. A test sensing pulse output means for generating a constant sensing pulse at a constant interval in response to a command from the center; and an automatic canceling means for automatically canceling the generation of the sensing pulse in the test sensing pulse output means. An image processing type vehicle detection device characterized by: 3. An image of a vehicle on a road is taken to give a bird's eye view, image processing is performed on this image data, the number of vehicles and the occupancy rate are measured, and the brightness of the original image data in the sensing measurement area is measured. A daytime presence detection signal holding method, which holds a presence detection signal whose distribution is ternary using an average luminance value of a background. 4. A background image data in a sensing measurement area by image-processing image data of a video camera and a video camera for taking a bird's-eye view of the movement of a vehicle on a road to measure the number of vehicles and an occupancy rate. The method for holding a presence detection signal during the daytime, characterized in that the presence detection signal is held based on the difference between the most frequent luminance ratio of the luminance of the image and the most frequent luminance ratio of the original image data. 5. A threshold value for determining a vehicle type according to the brightness of a sensed vehicle, while taking an image of the movement of a vehicle on the road to perform an image processing of the image data to measure the number of vehicles and an occupancy rate. Daytime vehicle type identification method characterized by changing the. 6. An image of a vehicle on a road is taken to give a bird's eye view, image processing is performed on this image data to measure the number of vehicles and an occupancy rate, and an image having a change in frame difference is displayed before the image. A vehicle extraction method characterized by determining a change in frame difference for a pixel, determining whether or not to perform expansion processing behind this pixel, and expanding only the vehicle portion without performing expansion processing on a shadow portion. 7. The number of vehicles and the occupancy rate are obtained by taking a picture of the movement of vehicles on the road and processing this image data using a background difference and a frame difference with expansion processing that make only the positive component valid. A twilight vehicle extraction method characterized by measuring and extracting a vehicle with low brightness by enabling the negative component of the background difference only during twilight. 8. A method of taking a bird's-eye view of trends on the road,
This image data is processed using the background difference that makes only the positive component valid and the frame difference with expansion processing to measure the number of vehicles and the occupancy rate. A vehicle extraction method characterized in that a pixel is also checked, and only when the pixel has a change, the pixel is validated to reduce a noise component. 9. The number of vehicles and the occupancy rate are photographed so as to give a bird's-eye view of the movement of vehicles on the road, and this image data is processed using a background difference and a frame difference with expansion processing that validate only the plus component. A vehicle extraction method characterized by performing measurement, automatically measuring the position of a white line drawn on a road with previously created background image data, and excluding the white line portion in the measurement area. 10. The number of vehicles and the occupancy rate are photographed so as to give a bird's eye view of the movement of vehicles on the road, and this image data is processed using a background difference and a frame difference with expansion processing that validates only the positive component. While measuring, the lane is subdivided, and the most frequent luminance value of the original image data and background image data in the subdivided area and this ratio, the average luminance value and the ratio of the pixels that change with the frame difference, the lane A non-predetermined parking existence determining method, comprising: determining whether or not a non-predetermined parking vehicle exists. 11. The number of vehicles and the occupancy rate are photographed so as to give a bird's-eye view of the movement of vehicles on the road, and this image data is processed using a background difference and a frame difference with expansion processing that validate only the plus component. A night-time background image data updating method characterized in that the background image data is updated by measuring the amount of change in the background image data in the negative direction and making it larger than the amount of change in the positive component. 12. The number of vehicles and the occupancy rate are photographed so as to give a bird's-eye view of the movement of vehicles on the road, and this image data is processed using a background difference that makes only a positive component valid and a frame difference with expansion processing. A night vehicle extraction method, which comprises measuring and performing image processing on a result of obtaining a logical sum of a pixel at a sample point and a pixel adjacent to the sample point only at night to extract a vehicle. 13. An image of a vehicle on a road is taken to give a bird's eye view, and this image data is processed using a background difference that makes only a positive component valid and a frame difference with expansion processing to determine the number of vehicles and the occupancy rate. Holds the presence detection signal at night time, which holds the presence detection signal from the luminance distribution of the result of the logical sum of the pixel at the sample point and the pixel next to the sample point in the detection output area only at night Method. 14. The number of vehicles and the occupancy rate are photographed so as to take a bird's-eye view of the movement of vehicles on the road, and this image data is processed using a background difference that makes only the positive component valid and a frame difference with expansion processing. Along with the measurement, only at night, the presence detection signal is held from the difference between the luminance distribution of the result of obtaining the logical sum of the pixel at the sample point in the sensing output area and the pixel next to the sample point and the luminance distribution of the background image data. A method for holding a night presence detection signal, comprising: