JPH0582632B2 - - Google Patents

Description

[Detailed Description of the Invention] (1) Industrial Application Field The present invention relates to a traffic flow measurement method for moving vehicles. This method uses the fact that the brightness value of the road surface is stable over time to recognize the road surface, and performs vehicle recognition based on changes in the brightness value.

(2) Conventional technology Information gathering and processing equipment using a traffic flow sensor composed of a group of minute photoelectric elements in order to spatially understand traffic flow phenomena is described in Mechanical Technology Research Institute Bulletin No. 35.
(1981), pages 22 (1st report), 83 (2nd report), and the same magazine, page 141 (3rd report), have reports by Shigeta et al.

In addition, the results of an experiment in which the entire road is imaged with a television camera, the panoramic image is used as electrical information, and the spatial average speed is measured from the binarized image obtained by quantization processing are described in "Traffic flow using image processing methods." Distributed as a material of the ``Measurement Technology'' Study Group (June 21, 1982).

(3) Problems to be solved by the invention Traffic flow measurement based on vehicle image processing techniques is a promising method because it can be realized over a wide spatial area rather than at a single point, and it can be easily transferred to online processing. However, as environmental conditions change, the brightness of vehicles and the road they drive on changes due to the weather, shadows from walls, etc., and changes in the amount of light projected by lighting, making it difficult to identify vehicles. This problem remains unsolved. .

(4) Means for solving the problems In order to solve the above problems, the method of the present invention constantly uses
The road surface reference brightness value is updated by changing the coefficient of the exponential smoothing method depending on the magnitude of the difference between the brightness value of the input image and the road surface reference brightness value, thereby providing feedback to the vehicle recognition algorithm.

This procedure is shown in FIG. The step in the figure is a step for determining the road surface brightness reference value, and is a procedure that is always executed before the step for measuring traffic flow, and the data for vehicle recognition is reviewed at each measurement time. There will be.

This cycle is shown in FIG. 4.

As shown in Fig. 1 and Fig. 2, m (plural) measurement lines that cross the running route are set in the input image, and one of the measurement lines A-A' is placed on the A-A' line. Time t at point (i, j)
Assuming that the brightness value at is I _i,j (t), the road surface reference brightness value is determined by the following method. That is,
In Fig. 1, it is assumed that the vehicle is traveling along the travel paths 2 and 2' in the direction of the arrow, and the value I _i,j (t) of the luminance value at the sample point (i, j) at time t is measured. . m sample points are provided in the direction of travel of the vehicle and n sample points are provided in the transverse direction, and the transverse direction is defined as the measurement line of the present invention. In Fig. 2, the luminance value of the sample point (i, j) of the input image at time t is I _i,j (t), and the road surface reference value at the sample point (i, j) at time (t-1) is I i,j (t). is expressed as I^ _i,j (t-1), and this value is updated at the measurement cycle to follow the surrounding measurement environment. In this case, the input luminance value I _{i,j (t) at the sample point (i,} j) at time t
The difference value S i _{,j (t) between the road surface reference luminance value I^ i, j} (t-1) is used, and the coefficient of exponential smoothing is changed depending on the magnitude of this value.

(5) Effect Exponential smoothing using a coefficient that changes the road surface reference brightness value by the difference value from the input image independently for each measurement period based on the procedure shown in Reference Figures 1 and 2 above. If vehicle recognition is performed while updating according to the law, it is possible to perform measurement processing that responds to changes in the environment.Also, if there is a sudden change in the environment in a certain measurement cycle, such as a vehicle accident or a vehicle stoppage due to traffic jams, measurement may fail once. Even if this is the case, vehicle recognition can be performed with high reliability using brightness values in other measurement cycles.

(6) Examples The process of updating the road surface reference brightness value will be described in detail.

For each sample point, from the current luminance value (I _i,j (t)) and the road surface reference luminance value (I^ _i,j (t−1)), the road surface reference value is calculated as follows using the exponential smoothing method. Update the brightness value.

The coefficient β _o used in the exponential smoothing method is changed depending on the difference value between the current brightness value and the road surface reference brightness value. An example is shown below.

I^ _ij (t)=(1−β _o )I^ _ij (t−1)+β _o I _ij (t), ) ｜I _i,j (t)−I^ _i,j (t−1) If |≦α ₁ , the coefficient β _o is set to 1.

) If α ₁ <|I _i,j (t)−I^ _i,j (t−1)|≦α ₂ , then the coefficient β _o is set to β ₁ .

) If α ₃ <|I _i,j (t)−I^ _i,j (t−1)|≦α ₄ , the coefficient β _o is set to β ₂ .

) If α ₄ <|I _i,j (t)−I^ _i,j (t−1)|, let the coefficient β _o be the coefficient β ₃ .

Furthermore, if the coefficient β _o is known in advance from the environmental conditions of the measurement area, that value may be applied to change the road surface reference brightness value.

(7) Effects If traffic flow is measured according to the method of the present invention, which is characterized by a method of tracking the road surface reference brightness value, vehicles can be reliably captured from the input image without being influenced by the measurement environment, and vehicle speed can be calculated online. , traffic flow parameters such as traffic volume can be known.

This effect is particularly effective when shadows are formed during the day, and when the brightness changes due to lighting being turned on or off, such as at night or inside a tunnel.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the relationship between measurement lines and sample points on an input screen in the present invention. FIG. 2 is a diagram for explaining the relationship between the road surface brightness reference value and the screen input brightness value, and the abbreviations are as follows. m: number of sample points in the road traveling direction, n: number of sample points in the road cross direction, the cross direction is called a measurement line. I _i,j (t): At time t - sample point of input image (i,
I^ _i,j (t-1): Road surface reference value of sample point (i, j) at (t-1), this value is updated in unit time and follows the environment. . S _i,j
(t): This is the difference value between the input luminance value at the sample point (i, j) at time t and the road surface reference value at time (t-1). FIG. 3 is a diagram showing a procedure constituting the present invention, and FIG. 4 is a diagram showing a specific measurement cycle in the present invention.

Claims (1)

[Scope of Claims] 1. In a traffic flow measurement method that recognizes the driving road surface with a surveillance camera and measures parameters such as the presence of vehicles and vehicle speed based on changes in the brightness value of the recognition screen, input at a certain time t. Calculate the difference value S _i,j (t) between the image brightness value I _i,j (t) and the road surface reference brightness value I^ _i,j (t-1) one measurement period ago, and calculate the exponential smoothing coefficient β _o is determined, and using the determined coefficient β _o , the road surface reference brightness value I^ _i,j (t) at time t is calculated using the following formula I^ _i,j (t) = (1 - β _o ) I^ _i,j (t-1) +β _o I _i,j (t) A method for updating a road surface reference brightness value in a traffic flow measurement method. 2. The method for updating a road surface reference brightness value according to claim 1, wherein the exponential smoothing coefficient β _o is determined according to the magnitude of the difference value S _i,j (t).