JPH03286399A - Image processing type traffic flow measuring device - Google Patents

Abstract

PURPOSE:To improve accuracy for the discrimination of vehicle kind and that for measuring the number of vehicles by detecting the shadow of the vehicle by utilizing nature that the density of the shadow is weakened as being separated from the vehicle. CONSTITUTION:The distribution density pattern of a differential value between the luminance of an input image image-picked up from the upper side of the vehicle targeted to be measured and road reference luminance changing sequentially against the advancing direction or the road traversing direction of the vehicle is detected as the front shadow or the side shadow of the vehicle by utilizing the nature that the density of both the front and side shadows of the vehicle are weakened as being separated from the vehicle. In other words, an image ternarization processing part 23 performs ternarization(1, 0, -1) at every measuring sampling point by using a luminance signal, and a vehicle detecting part 24 detects the vehicle from the deviation of the luminance at the measuring sampling point. A shadow discrimination part 25 discriminates the shadow of the vehicle based on the distribution pattern of the differential value between the luminance value of the input image and the road reference luminance. A vehicle measuring part 26 identifies the vehicle by using the shadow data of the vehicle obtained from the shadow discrimination part 25, and calculates every kind of traffic flow measuring data without being affected by the shadow. In such a way, it is possible to improve the accuracy for the discrimination of the vehicle kind and that for measuring the number of vehicles.

Description

[Detailed Description of the Invention] [Industrial Application Field 1] The present invention relates to an image processing type traffic flow measuring device that detects vehicles through image processing using an industrial television (ITV) camera, etc. This invention relates to a traffic flow measurement device that can detect shadows that occur around a vehicle with high precision and remove the effects of shadows.

[Prior art] In this type of traffic flow measuring device, the dark blue that appears around the vehicle may be caused by incorrectly measuring the size of the vehicle, erroneously detecting the type of vehicle, or mistaking it for another vehicle. This may cause the number of vehicles to be measured to be erroneously detected. Therefore, it is necessary to accurately detect the shadows that appear around the vehicle and remove the influence of the shape.

The conventional method for detecting the shape of a vehicle in this type of traffic flow measuring device is to detect the shadow by taking advantage of the fact that the shadow of the vehicle is darker than the road surface, and the brightness is uniform with almost no change over time. A method is generally known (see Hashimoto, Kumagai et al., "Development of Image Processing Traffic Flow Measurement System", Sumitomo Electric No. 127, P59, September 1986).

[Problem to be Solved by the Invention 1] However, as mentioned above, the conventional technology does not utilize the fact that the brightness of the shadow area is darker than the road surface or that the temporal change in brightness value at the measurement point is uniform. Because shadows were detected using the 3D image, it was difficult to distinguish the vehicle from a black vehicle whose brightness value was lower than the road surface standard brightness.

An object of the present invention is to solve the above-mentioned problems and provide an image processing type traffic flow measurement device that can accurately remove the influence of shadows and improve the accuracy of vehicle type discrimination and single-unit measurement.

[Means for Solving the Problems 1] In order to achieve the above-mentioned object, the present invention detects a portion having a different brightness from the road surface portion from an input image obtained by photographing a road surface and a vehicle existing on the road surface. In a device that detects the vehicle by extracting it as a feature point,
a calculation means for calculating a difference value between a brightness value of the input image having a brightness lower than a reference brightness of the road surface portion and a reference brightness value of the road surface portion; and a distribution pattern of the difference value calculated by the calculation means; The vehicle is characterized by comprising a shadow detection means for detecting a portion of the road surface that changes sequentially along the longitudinal direction or the transverse direction as a moving portion of the vehicle.

[Function 1] In the present invention, focusing on the characteristic that the shadow of a vehicle becomes thinner as it moves away from the vehicle body, the distribution density pattern of the difference value between the brightness of the input image captured from above the vehicle to be measured and the road surface reference brightness is developed. The area where the shadow changes sequentially in a certain direction is detected as the shadow of the vehicle.

In this way, the present invention detects the shadow of a vehicle by taking advantage of the property that it becomes thinner as it moves away from the vehicle body, so it is possible to stably detect the shadow regardless of the body color or shape of moving or stationary vehicles. Even if the shadow extends over the lane, it will not be mistakenly detected as a black vehicle. Therefore, according to the present invention, the influence of shadows can be effectively removed, and the accuracy of vehicle type discrimination and five vehicle count measurement accuracy can be improved.

[Example] Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 1 shows the main circuit configuration of an image processing type traffic flow measuring device according to an embodiment of the present invention. In this figure, 1 is a camera unit (fixed camera) such as an I-D camera as an imaging means for capturing an image of a vehicle traveling on a road, and 2 is a camera unit that processes the output image signal of the camera unit 1 to capture images of the surroundings of the vehicle. This is the imager θ1j section that detects shadows that occur in the image. Image measurement section 2 is A/D
(analog/digital) conversion section 211 image storage section 221
Image ternarization processing unit 23. Vehicle body detection section 24. Shadow determination section 25
．． It is composed of a vehicle measuring section 26 and an output section 27.

The A/D conversion unit 21 digitizes the output image signal of the camera unit 1. The 1-image storage unit 22 is comprised of a RAM (random access memory), etc., and stores the digitized image signal as a luminance signal in units of imaging screens. . Image ternarization processing 23 performs ternarization (1, O, -1) for each measurement sample point using the stored luminance signal. The vehicle body detection unit 24 detects the vehicle body from the luminance shift of the measurement sample points. The shadow determination unit 25 determines the shadow of the vehicle as described below based on the distribution pattern of the difference value between the brightness value of the input image and the road surface reference brightness value.

The vehicle measurement unit 26 identifies the vehicle using the vehicle shadow data obtained from the shadow determination unit 25, calculates various traffic flow measurement data that is not affected by shadows, and outputs this measurement data from the output unit 27 to the remote traffic flow. Send to solid control device.

FIG. 2 shows an example of how the camera section 1 and image measurement section 2 shown in FIG. 1 are actually installed. Camera part 1 is multi-lane road 4
The image measurement unit 2 is fixed to the upper arm part of a pillar (ball) 3 erected on the roadside 5 of , and the image measurement unit 2 is fixed at a predetermined position on the trunk of the pillar 3. The camera unit 1 images a predetermined range of the multi-lane road 4 from a certain height at a certain angle as a measurement area 6, and the image measurement unit 2 performs vehicle sensing processing from the image data captured by the camera unit 1. In this embodiment, a case where there are three lanes on each side will be considered as an example. Note that 7 indicates the center line (median strip).

FIG. 3 shows the principle of the determination operation in the shadow determination section 25 described above. As shown in Figure 3 (A), vehicle body shadows can be divided into front shadows and horizontal shadows depending on where they occur, but since the length or width of each is equal to the length or width of the vehicle, they are different from the vehicle in terms of shape. It is difficult to separate the shadows. However, in Figure 3 (
As shown in (B) and (C) of the same figure, the shadow of the vehicle body is the front shadow,
Both sides become thinner as they move away from the car body. The shadow determination unit 25 utilizes this phenomenon to detect the shadow of the vehicle body in the following manner. This detection procedure is the same for both foreshadow and horizontal shapes, so the foreshadow case will be described as an example.

First, a measurement line is created by connecting each measurement sample point in the vehicle traveling direction and the road crossing direction. Therefore, the measurement points are arranged in a grid pattern in the measurement area 4. In the case of a front shadow, the measurement line (a) in the vehicle traveling direction is used. A portion where the input luminance is darker than the road surface reference luminance in a certain continuous section on the measurement line in the vehicle traveling direction is detected. The difference value between the road surface reference brightness and the input brightness is calculated for each measurement sample point included in the detected section. For example, when the measurement line (a) is scanned from the downstream side of the road toward the upstream side, the slope with the above difference value is calculated. If the value gradually increases with , it is extracted as a shadow candidate.

Next, the measurement lines are shifted one by one in the direction of the road cross section and the same process as above is repeated. If a shadow candidate exists across multiple measurement lines and is extracted as a plane area, this area is used as a shadow (front). Detected as a shadow).

Further, the operation of the embodiment of the present invention will be explained in detail with reference to the flowchart of FIG.

The analog image signal from the camera section l is sent to the A/D converter section 2.
1, it is converted into digital luminance data of, for example, 256 gradations, and is recorded in the image storage unit 22 on a screen-by-screen basis (step SL). After the luminance data for each pixel at the measurement point is extracted by pixel extraction, the difference (difference value) from the road surface reference luminance is calculated for each pixel by luminance calculation (step S2). The difference value data for each pixel is processed by r in the image ternarization processing unit 23.
+lJ (brighter than the road surface), rOJ (same as the road surface)
, r-1 (darker than the road surface) (step S3). That is, the road surface normally has a short, almost constant brightness, and this brightness (road surface reference brightness) does not change over a certain short time interval, and is fixed depending on the measurement area. Therefore, this reference brightness is determined in advance or at any time, and compared with the brightness data of each measurement point. Then, for example, a pixel whose luminance is higher than the road surface by a certain level...+1 a pixel whose luminance is almost the same as that of the road surface...0 a pixel whose luminance is lower than the road surface by a certain level or more...-1 is converted into three values.

Next, it is detected whether the ternary data of each pixel is darker than the road surface. (Step S4) That is, in the case of this embodiment, since the ternary data of the road surface part is "0", the part where the ternary data is "-l" is detected. Become. Then, check whether the dark areas detected in this way are continuous in the vertical and horizontal directions (
Step S5) ,,Subsequently, a measurement line (al
is scanned, and the difference value between the road surface reference brightness and the input brightness performed for each measurement sample point gradually increases with a certain slope from downstream to upstream of the road, which is recognized as a vehicle foreshadow ( Step S6).

Here, assuming that the vehicle is running on the road in the direction of camera unit 1 or is stopped, the hood will always appear behind the front shadow, and the roof will always appear behind the hood. It is something. Also,
The brightness of the hood and roof is higher than the brightness of the front shadow, and is often higher than the brightness of the road surface. However, its brightness is not constant compared to the brightness of the road surface. Therefore, the bonnet and roof of the vehicle can be detected by checking whether there is a pattern with the highest brightness next to the foreshadow or a pattern with disordered brightness. These detection results are sent to the next vehicle measurement unit 26 (step S7), and it is determined whether the front shadow, bonnet, and roof of the detected vehicle appear within a certain range in the lateral direction. is done (step S
8). That is, the front shadow of each vehicle as shown in FIG. 3(A).

Since the hood and roof should each appear within a certain range in the lateral direction, they can be recognized as one vehicle (step 511).

On the other hand, if the width of the bonnet or roof is not within a certain range, this generally means that the vehicle has a horizontal shape, so check to see if there are parts on the side of the vehicle that are darker than the road surface and are continuous in both the vertical and horizontal directions. Then, the measurement line (b) is scanned toward the road side 5 or the center line 7 of the continuous dark part, and the road surface reference brightness and the road surface reference brightness performed for each measurement sample point are examined. If the difference value of the input luminance gradually decreases (or increases) with a certain slope from the road side 5 toward the center line 7, it is recognized as the horizontal shape of the vehicle.In step 51op, based on the above measurement results, It is recognized as a vehicle (step 511).

After the above processing is completed, the vehicle measurement unit 26 performs vehicle tracking, lane determination, vehicle measurement, etc. and,
The measurement results are sent from the output unit 27 to a remote solid device for traffic control and are output and displayed.

That is, the vehicle measuring section 26 uses the shadow of the vehicle detected by the shape determining section 25 as noise, removes the shadow portion from the data in the image storage section 22, and recognizes only the vehicle body.

This vehicle body recognition is performed, for example, by detecting the rise and fall of vehicle data within the measurement area. That is, the measurement area is divided into an upstream area and a downstream area, and the traffic volume is counted as passing vehicles at the start of vehicle data in the downstream area.
In addition, the length of the vehicle is measured from the distance between the rise of the vehicle data and the fall of the vehicle data in the downstream area, and the type of vehicle is determined as large or small.Furthermore, the speed of the vehicle (vehicle speed) is calculated from the travel distance and travel time of the vehicle. do. These traffic flow measurement data are sent from the output unit 27 to a remote solid device.

Therefore, the image measurement unit 2 can obtain accurate vehicle type discrimination data, passing vehicle number measurement data, etc. that are not affected by shadows.

Note that the present invention is not limited to the above embodiments, and various modifications are possible.

In the embodiment, the shadow under the bumper of the vehicle is detected, but the invention is not limited to this. For example, a moving vehicle may be photographed from the rear using a camera, and shapes formed behind the vehicle may be detected. Furthermore, the vehicle is not limited to one that runs on a road surface, but may be one that is parked on the roadside or in a parking lot.

As the camera 1, for example, a CCTV camera can be used, but it is not limited to this, and it is not essential to use the image storage section 22.

In other words, it is also possible to directly connect the camera and the image processing device and perform online real-time processing.

Luminance data is not limited to 2'=256 gradations, but can also be
It is also possible to use a different tone such as '=16 gradations, 2'=64 IN tones, etc. Furthermore, the luminance data is not limited to ternary data, and may be any data such as binarized, quaternary, or quinarized data.

[Effects of the Invention J] As explained above, according to the present invention, the input image captured from above the vehicle to be measured takes advantage of the property that the shadow of the vehicle becomes thinner as it moves away from the vehicle body, both in front and sideways. The difference value distribution density pattern between the luminance of Shadows can be detected stably and accurately regardless of the vehicle type, improving the accuracy of vehicle type determination (large/small vehicle type discrimination), and even when shadows extend onto adjacent lanes, they can be mistakenly detected as black vehicles. Since there is nothing to do, it is possible to improve the accuracy of traffic volume measurement.

In addition, the present invention detects vehicle-shaped parts across multiple lanes, and uses this to identify individual vehicles, while also recognizing vehicles in combination with protruding parts such as the bonnet. , or even if there is a vehicle traveling while changing lanes, or if two or more vehicles are running parallel to each other in multiple lanes, these can be accurately recognized.

3... Prop, 21...A/D conversion section, 22... image storage unit, 23... Image ternarization processing unit, 24...Vehicle body detection section, 25...Shadow determination section, 26...Vehicle measurement section, 27...Output section.

[Brief explanation of the drawing]

Fig. 1 is a block diagram showing the circuit configuration of an embodiment of the present invention, Fig. 2 is a perspective view showing an installation example of the device shown in Fig. 1, and Fig. 3 (
A), (B) and (C) are explanatory diagrams illustrating the nature of the shape of a vehicle and the operation of determining its shadow. FIG. 4 is a flowchart showing the operation details of the embodiment of the present invention.

Claims (1)

[Claims] 1) From an input image obtained by photographing a road surface and a vehicle existing on this road surface, the vehicle is detected by extracting a portion having a different brightness from the road surface portion as a feature point. In the apparatus, a calculation means for calculating a difference value between a brightness value of the input image having a brightness lower than a reference brightness of the road surface portion and a reference brightness value of the road surface portion, and a distribution of the difference value calculated by the calculation means. An image processing type traffic flow measuring device comprising: shadow detection means for detecting a portion where a pattern changes sequentially along the longitudinal direction or the transverse direction of the road surface as a shadow portion of the vehicle.