JPH06301884A - Automatic image recognizing device - Google Patents

Abstract

PURPOSE:To prevent the brightness of a road surface from being erroneously recognized and the non-operation of fuzzy diaphragm opening control from occurring by eliminating a sampled painted part from a road surface image. CONSTITUTION:A recoginiton processing part 9 finds mean luminance and a degree of counter light, etc., in addition to character recognition, and is provided with a function to control the diaphragm opening of a camera 4. Also, it is provided with an elimination function to a painted part to accurately calculate the mean luminance, etc. In other words, the histogram of an image fetched by the camera 4 is generated, and when the painted parts in white and yellow occupy >=5% in the whole, the periphery of the white and yellow parts is magnified by one picture element. When picture elements >=100 in a cluster of picture element groups of white and yellow parts exist, the cluster is decided as the painted part, and the cluster is set as an (area outside target) for following calculation processing. After that, a picture element group area decided as the one outside target is managed by, for example, a plane for masking, and the picture element group area(i.e., the painted part) is eliminated from a screen.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to automatic image recognition of a license plate automatic recognition device or the like configured to automatically extract a paint portion from a road surface image and exclude the paint portion when calculating road surface information. The present invention relates to a device (AVI device).

[0002]

2. Description of the Related Art On a highway or the like, a license plate of a running vehicle may be automatically read in order to collect various traffic data. Therefore, the vehicle running on the vehicle is read every predetermined period. An automatic license plate recognition device is used which recognizes a license plate of a vehicle by photographing and processing the obtained image. In this type of license plate automatic recognition device, there is a camera aperture that performs fuzzy aperture control by using the backlight intensity as an input function because it is necessary to automatically adjust the aperture appropriately according to the degree of backlight, forward light, weather, and the like. In this case, in order to obtain road surface information such as the average brightness and variation of the road surface, it is calculated from the brightness of pixels at sample points that are uniformly scattered in the screen.

[0003]

However, in the conventional method of obtaining the average brightness and the variation as described above,
If there is a white or yellow paint display on the road surface, the brightness of the pixels changes, and the average brightness and variations increase. As a result, when the average brightness becomes high, the brightness of the road surface is erroneously recognized, or if the variation becomes large, it is determined that the road surface is not “good” and the fuzzy aperture control does not operate. .

Therefore, the present invention has been made in view of the above problems, and an object of the present invention is to provide an automatic image recognition apparatus which solves erroneous recognition of road surface brightness and malfunction of fuzzy aperture control. To do.

[0005]

In order to achieve the above object, the automatic image recognition apparatus of the present invention comprises a photographing means for photographing a road surface and a road surface information calculating means for calculating road surface information from the photographed image. In the automatic image recognition device, a histogram creating means for creating a histogram of the road surface image obtained by the photographing means, a paint portion extracting means for finding a paint portion of the road surface from the histogram, and an extracted paint portion from the road surface image. And a paint portion removing means for removing the paint portion.

[0006]

In this automatic image recognition apparatus, the histogram creating means creates a histogram of the road surface image obtained by the photographing means, and the paint portion extracting means draws white or yellow on the road surface in this histogram. The paint portion is indexed, and the paint portion removing means removes the extracted paint portion from the road surface image. Therefore, when calculating the road surface information (average brightness, variation, etc.) by the road surface information calculating means, the paint portion is excluded from the target,
The increase in the average brightness does not cause the brightness of the road surface to be erroneously recognized, and the fuzzy aperture control operates reliably.

[0007]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An AVI device of the present invention will be described below based on embodiments. FIG. 2 is a perspective view showing the outline of a system in which the present invention is implemented. A main body 3 of the automatic license plate recognition device is installed on a pillar 2 provided on the road side of the traveling road 1, and a camera 4 and an illuminating lamp 5 are provided on a horizontal rod of the pillar 2. The camera 4 and the main body 3 are electrically connected. When the vehicle 6 travels on the traveling road 1, the front part of the vehicle 6 captured by the camera 4 is photographed and the license plate is read.

FIG. 1 is a schematic block diagram of a license plate automatic recognition apparatus showing an embodiment of the present invention. The license plate reading device stores an image processing unit 7 that performs a predetermined process such as normalizing an image captured by the camera 4 at each sampling time (predetermined time) and a normalized image. An image memory 8 and a recognition processing unit 9 for recognizing the license plate by performing various processing such as cutting out the license plate portion based on the data stored in the image memory 8. Recognition processing unit 9
Then, in addition to the function of recognizing characters, it has a function of controlling the aperture of the camera 4 by obtaining the average brightness of the screen, the degree of backlight, and the like. Further, in order to accurately calculate the average brightness and the like, a function of removing the paint portion is also provided. However,
Other than the removal processing of the paint portion of the road surface, the present invention is the same as the conventional AVI device, and the present invention is characterized by the algorithm of how to recognize and remove the paint portion of the road surface. The procedure will be mainly described.

FIG. 3 shows a screen of an AVI device on a road surface including a paint portion where "large" 11 and "↓" 12 are displayed. Further, FIG. 4 shows a screen when the paint portions 11 and 12 are set as the non-processing target areas by the following processing procedure. Specific processing will be described with reference to the flowchart shown in FIG. First, in step (hereinafter abbreviated as ST) 1, a histogram of a road surface image captured by a photographing means (for example, a camera) is created. This histogram is
For example, as shown in FIG. 6, the road surface portion A having no paint portion has a large number of pixels in the dark portion, and the paint portion B of white or yellow has a small pixel number in the light portion. In such a histogram, it is determined whether the white / yellow paint portion B is within 5% of the whole (ST2).
Here, the numerical value of 5% is an empirical value of average paint in the visual field of the 4-digit AVI device, and may be appropriately changed depending on the visual field.

If the answer is NO in ST2 (when the paint portion B exceeds 5%), it is determined in ST4 that the white / yellow portion is not the paint portion, that is, reflected light or vehicle. If YES, the periphery of the white / yellow portion is enlarged by one pixel (ST3). This pixel enlargement is to take into account the screen shake. In ST5, it is determined whether or not the number of pixels in the group of pixels in the white / yellow portion is 100 or more. If the number of pixels is less than 100, the group of pixels in the group of white / yellow is determined to be noise. . On the other hand, when the number of pixels is 100 or more, it is determined that the lump is the painted portion, and the lump is set as the "non-target area" of the subsequent calculation processing (ST6). If there are a plurality of pixel groups, the process is performed for each group.
As for the number of pixels 100, if the dents and the roughness of the road surface are excluded from the target area, the target area may be too large. The noise is less than several hundreds. Of course, the number of pixels may be changed appropriately.

After that, the pixel group area determined to be out of the object is managed by, for example, a mask plane (ST8), and the pixel group area (that is, the paint portion) is excluded from the screen (ST9). As shown in FIG. 5, a screen in which the paint portions 11 and 12 are the non-processing target areas is obtained. Then, the average luminance, the in-screen variance value (variation), etc. are calculated from the road surface image as in the conventional case (ST10).

The above algorithm is operated periodically (for example, every 5 minutes) to sequentially update the pixel group area managed by the mask plane or the like.

[0013]

Since the AVI device of the present invention is constructed as described above, it has the following effects. (1) Since the paint part is excluded and excluded from the calculation process of the road surface information, the average brightness of the road surface not including the paint part can be calculated, and the brightness of the road surface is erroneously recognized. Disappears. (2) Since the paint portion is excluded and excluded from the road surface information calculation processing, it is possible to calculate the road surface variation that does not include the paint portion in the calculation, and the road surface variation is reduced. (3) The reliability of the secondary information using the average brightness and variation of the road surface is improved. (4) Even if there is a paint part on the road surface, the fuzzy aperture control operates reliably. (5) When installing a camera (imaging means) for capturing an image of the road surface, it is not necessary to consider that there is no paint portion on the road surface in the road surface installation conditions, and the installation conditions are reduced accordingly.
Fewer restrictions on the installation location. (6) Even if paint is displayed on the road surface after the camera is installed, the painted portion is automatically removed, so maintenance is easy.

[Brief description of drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of an automatic license plate recognition apparatus according to an embodiment of the present invention.

FIG. 2 is a perspective view showing the outline of a system in which the present invention is implemented.

FIG. 3 is an image diagram of a road surface before the paint portion of the road surface is removed in the automatic license plate recognizing apparatus according to the embodiment.

FIG. 4 is a road surface image diagram after the paint portion of the road surface is excluded in the automatic license plate recognition device.

FIG. 5 is a flowchart for explaining an algorithm of a paint portion removal process in the automatic license plate recognition apparatus.

FIG. 6 is a diagram showing a histogram of a road surface image created by a paint portion removal process in the automatic license plate recognition apparatus.

Claims (1)

[Claims] 1. An automatic image recognition apparatus comprising a photographing means for photographing a road surface and a road surface information calculating means for calculating road surface information from the photographed image, wherein a histogram for creating a histogram of the road surface image obtained by the photographing means. Creating means, and paint part extracting means for calculating the paint part of the road surface from this histogram,
An automatic image recognition device, comprising: a paint portion removing unit that removes the extracted paint portion from the road surface image.