JPH0336698A - Parking area congestion state deciding device - Google Patents

Abstract

PURPOSE:To obtain an accurate binary screen by dividing a screen into small rectangular areas, calculating the brightness of a road surface, calculating an inter-class dispersion value, and then calculating a binary threshold value. CONSTITUTION:Each rectangular are R(i, j) 1 is constituted of p X q picture elements and a screen 2 is constituted of n X m rectangular areas R(i, j) 1. A threshold value KE(i, j) is set up by calculating a dispersion value B(i, j) from a brightness histogram in the divided areas R (i, j) 1 and the brightness of the road surface is calculated from the histogram of the dispersion value B(i, j). A binary value is selected and obtained in accordance with the size of the inter-class dispersion value. Consequently, a highly accurate paring area congestion state deciding device capable of obtaining an accurate background screen can be obtained even in a parking area always including parked vehicles.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a parking lot congestion state determination device used in parking lots provided at expressways, toll roads, event venues, etc.

(Prior art) The binary screen creation method used in conventional parking lot congestion state determination devices updates the created background image with time and calculates the screen difference between the background image and the newly captured image. The differential signal is then binarized using a predetermined threshold. In this way, even with the above binary screen creation method, a binary screen can be obtained by creating a background screen and performing screen difference.

(Problem to be Solved by the Invention) However, in the method of creating a background screen while removing moving objects as in the above configuration, when vehicles are parked for a long time such as in a parking lot, Because the background screen includes the following, it is not possible to create a background screen with no parked vehicles at all.

There was a problem that an appropriate binary screen could not be obtained.

The present invention solves the above problems and provides a parking lot congestion state determination device that can obtain a binary screen with high accuracy.

(Means for Solving the Problems) In order to solve the above problems, the present invention provides a road surface brightness calculation section and a binarization threshold calculation section.

(Function) With the above configuration, the road surface brightness calculation unit calculates the variance value within the divided rectangular area.

The road surface brightness is calculated, and the binarization threshold calculation unit can obtain an optimal threshold for creating a binary screen based on the calculated road surface brightness, thereby obtaining a highly accurate binary screen. be able to.

(Example) An example of the present invention will be described with reference to FIGS. 1 to 5.

FIG. 1 is a block diagram of a screen according to the present invention divided into small rectangular areas, where the rectangular area R(i,j)l is.

It is composed of pixels of p width x height q, and -plane 2 is composed of a rectangular area R(l l j) l of n width x height m. The number of screen divisions is determined by the installation height of the camera used for shooting.

This is uniquely determined by the angle of depression, focal length, etc.

FIG. 2 is a flowchart showing the processing procedure of the present invention. In "Calculation of variance value Bt (x-j)" 3, a histogram within the divided rectangular region R(i, j) 1 is created and the inter-class variance value is calculated (Journal of the Institute of Electronics and Communication Engineers).

Nobuji Otsu "Automatic threshold selection method based on discriminant and least squares criterion"'80/4 Voff, J63-
DNo., 4゜p349-356) L, the value when the variance value is maximum is set as the variance value Bx(1+j) in the rectangular area Rb l j) 1.

Subsequently, in "Threshold value Kst (L j) calculation" 4, the average luminance value at this time is set as the threshold value □(i,,i). The same rectangular area R (
For the luminance below the threshold □(i, j) at i, , iH, create a histogram in the same way as above, calculate the interclass variance values B, (i, j), and set the threshold K
Set lz(i, j〉. In "All area completion judgment" 7,
Variance value B(1*j) and threshold value K of all rectangular areas
Determine whether the calculation of l + (11j) has been completed, and if it has not been completed, "Calculation of variance value B (i, j) and threshold value (i, j)" 3,4.5 and repeat 6.

Next, in order to extract road surface candidates with a small luminance distribution within the divided rectangular regions, "reference threshold B" is set.

8, and the preset threshold value B.

, a region smaller than the threshold value B is extracted, and a histogram of K*1Cxe j) is created in "histogram creation" 9. Next, in "All data completion judgment" 10,
Determine whether data in all areas has been completed, and continue "histogram creation" 9 of Kmt (is j) until completion.
repeat. "For the road surface brightness, from the histogram of Kmtbtj) obtained in calculation J11, the lower limit percentage threshold PTL and the upper limit percentage threshold PT are calculated.
Using rI, remove peculiar data other than road surface candidates, calculate y for the average road surface brightness and □ for the road surface brightness standard deviation,
The road surface brightness is set as (Kll=KR+σXKIB). Note that the value of σ is uniquely determined by the distribution of road surface brightness.

``Comparison with standard threshold value TH, □'' 12 is the first variance value judgment unit of the divided rectangular area, which compares it with a preset threshold value THo, and if it is smaller than TH, then it is divided. ' T Hb l as the final binarization threshold in the area
j) = Set Kll''13. Threshold value TH
□In the above cases, the threshold value T preset in the comparison J14 with the r reference threshold value TE01 which is the second variance value determination section
When compared with H,, TH, is smaller.

As a threshold value r'rH(it j)=Kwt(is
j) 15L,, When TH is 112 or more, set □(itj) to rTH(i, j)= as the threshold J 1
6.

In "all area completion determination" 17, it is determined whether or not all areas have been processed, and if the processing has not been completed, the calculation of the binarization threshold for the divided areas is repeated.

Figure 3 is a road surface brightness distribution diagram showing an example of the histogram of a divided area and the inter-class variance at that time. The value Kmt(L j) will be obtained.

FIG. 4 shows the threshold value KIl□(11j
) Road surface brightness distribution map when the inter-class variance is calculated from the histogram in the area excluding the larger brightness.

Threshold value K a2C1l when the inter-class variance value is maximum between the maximum points MPI and MP2 of the histogram
j) is obtained.

FIG. 5 is a threshold distribution diagram showing an example of a histogram for calculating road surface brightness, in which the lower limit percentage threshold PT and the upper limit percentage threshold PTrl are used to remove area data other than road surface candidates. The area excluding the singular data is the shaded area.

(Effects of the Invention) As explained above, according to the present invention, by dividing the screen into small rectangular areas and extracting the small inter-class variance value of the luminance distribution within each divided area as the road surface area. , road surface brightness can be calculated. The binarization threshold value is selected based on the magnitude of the inter-class variance value, and the binarization threshold value is obtained with high accuracy, and an accurate background screen can be obtained even in a parking lot where there are always parked vehicles. A highly accurate parking lot congestion state determination device becomes possible.

[Brief explanation of drawings]

Figure 1 shows C of the parking 1ih congestion state determination device according to the present invention.
A plan view showing the division state of the RTR surface, Fig. 2 is a processing flowchart showing the procedure of the binarization threshold calculation process, and Fig. 3 is a road surface showing an example of the histogram of the divided area and the inter-class variance value at that time. Brightness distribution map. Figure 4 is a road surface brightness distribution diagram showing the histogram and inter-class variance when brightness greater than the threshold value Kmx (1 + J) of the divided areas is removed from Figure 3. Figure 5 is a road surface brightness distribution diagram. FIG. 7 is a threshold distribution diagram showing an example of a histogram when calculating brightness. 1... Rectangular area R(IIj), 2... Screen.

Claims (1)

[Claims] Divide the grayscale image into small rectangular areas, and set a threshold value K_s(i, j) by calculating the variance value B(i, j) from the brightness histogram in the divided rectangular areas;
The road surface brightness was calculated from the histogram of the calculated variance value B(i, j), and a binarization threshold within the rectangular area was set to remove noise from the road surface, shadows, etc. while preserving the characteristics of the vehicle. A parking lot congestion state determination device characterized by: