JPS61176808A - Vehicle identifying device - Google Patents

Abstract

PURPOSE:To improve the specification accuracy of a vehicle by providing a character recognizing part which recognizes the character of a number plate and a characteristic extracting part which extracts the characteristic of the vehicle among picked up images and a collating means which compares and collates the characteristic of the vehicle and the information on the vehicle kind and vehicle model stored by a storage means. CONSTITUTION:The image signal picked up by an image pickup part 11 is stored into a frame memory 12 subjected to an along-to-digital conversion and the number plate is indexed from the images by a number plate indexing part 13. The character in the number plate is indexed in a character indexing part 14 and is recognized in the character recognizing part 15. The character extracting part 17 extracts the characteristics such as the shape of the front grill and bumper and the shape of the lamp of an automobile by scanning the image in the frame memory 12. The characteristics extracted in the part 17 and the dictionary of the characteristics by each of the vehicle kind and model stored in a dictionary memory 19 are compared and collated in the collating part 18.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention automatically identifies a traveling vehicle such as a car,
The present invention relates to a vehicle identification device that identifies a vehicle, and particularly relates to a wheel identification device that recognizes number plates in an optically scanned image of the front or rear surface of a vehicle and identifies the vehicle type and chassis from the image.

[Conventional technology]

In general, in order to identify and specify a vehicle such as a moving car, it is sufficient to be able to identify the license plate and recognize the characters on the license plate, and when capturing an image of a moving vehicle, it is necessary to capture an image of at least the width of the vehicle. is necessary.

However, no Maki wheel identification device has ever been put to practical use.

[Problems to be solved]

When identifying wheels as mentioned above, if the license plate has been used for a long time, the text may be unclear, or the contrast with the background may be difficult for machines to recognize due to scratches or dirt. , and when capturing an image of a moving car with a commonly used camera,
There was a problem in that the characters on license plates could not be recognized with high accuracy and sufficient resolution could not be obtained.

SUMMARY OF THE INVENTION The present invention solves the above-mentioned problems, and aims to provide a vehicle identification device that improves the accuracy of identifying a vehicle.

[Means for solving problems]

The configuration of the vehicle identification device according to the present invention for achieving the above object of the present invention includes an imaging means that optically scans and images the front or rear surface of a vehicle such as a running automobile, and an imaging means that takes an image using the imaging means. a license plate indexing unit that extracts the license plate of the vehicle from the image obtained by the imaging means; a character recognition unit that recognizes the characters on the license plate; and a feature extraction unit that extracts the characteristics of the vehicle from the image obtained from the imaging means. , and a collation means for comparing and collating the vehicle characteristics obtained from the feature extraction section with the vehicle type and vehicle type information stored in the storage means.

〔Example〕

Next, one embodiment of the present invention will be described based on the drawings.

FIG. 1 is a block diagram showing an embodiment of a vehicle identification device according to the present invention.

FIG. 2 is a partial front view showing the front of the vehicle.

FIG. 3 is an image diagram scanned by the imaging section in the vehicle identification device according to the present invention.

FIG. 4 is a block diagram of the feature extraction section in the same embodiment.

In the figure, reference numeral 11 is composed of a normal vidicon or COD camera and a Doppler type or loop coil type vehicle detector, and serves as an imaging means for photographing, for example, the front side of a wheel detected by the vehicle detector. Department. The image signal captured by the image capturing unit 11 is A/D converted and stored in the frame memory 12, and the nano (
- Index the plate 50 by the license plate indexer L3.

Reference numeral 4 denotes a character indexing section by which the characters in the license plate 50 are indexed and recognized by the character recognition section 15. In addition, the character indexing section L4 and the character recognition section 15 are as follows:
A conventional optical reader (OCR) is applied.

17C is a feature extraction unit that extracts features of parts other than the license plate 50 in the image of the vehicle stored in the frame memory. Feature extraction part C, vehicle width detection sensor'I
It is determined based on the vehicle width information detected by ``16'' and the license plate position information obtained by the license plate indexing section 13, and the characteristics are extracted by scanning the image in the frame memory 12. It is something to do.

The vehicle width detection upper sensor 16 is configured with a photodiode array placed at a gate on the road, a pressure sensor placed on the road, or the like, and detects the width of a traveling vehicle.

Reference numeral 18 denotes a collation unit that compares and collates the nine features extracted by the feature extraction unit 17 with a feature dictionary for each vehicle type and vehicle quantity stored in the dictionary memory 19. Further, 20 is registered in advance for supplementation using the license plate character recognition result obtained by the character recognition unit 15 and the vehicle type and table type identification result obtained by the collation unit 18. This is a search unit that performs a search by matching with a vehicle.

The collation section 18 and the dictionary memo IJ 19 are constituted by ordinary microcomputers, and the search section uses a mini combiator and is processed by software.

On the other hand, the features extracted by the feature extraction section include the front grille 52 of the automobile, the shape of the -54 mark, the shape of the lamps, etc. In this embodiment, as an example of displaying these features, As shown in FIG. 4, several areas are defined based on the license plate position, and the distribution of the number of black and white change points, the number of continuous black points, and the number of white points when each area is scanned left and right, and up and down is shown.

Further, FIG. 3 shows an example of feature extraction obtained in each of the above regions, which was performed on a part of the front grill of an automobile. 31 in the figure is an area for extracting features in the captured image, and vertical scanning 31a1 is indicated by the broken line in the figure.
The horizontal direction scan 31b is scanned. 32 and 33 are the frequency distributions of points changing from black to white obtained for each vertical and horizontal scan, respectively, and 35 is the number of consecutive white and black points obtained for each scan, that is, the number of black points among the black points. , is the frequency distribution for each number of white dots.

Here, the license plate indexing section L3 is performed by utilizing the fact that the characters in the license plate 50 are arranged according to a predetermined arrangement as shown in FIG. That is, the number plate 50 was photographed within the nine-frame area.
Extract the circumscribed rectangle of the black dot arm (white dot cluster when the background part of the license plate is back or black) corresponding to the size of the characters inside,
The license plate 50 is discovered based on its size and arrangement. This method can apply the character indexing method used in ordinary optical character reading devices.

Next, based on FIG. 4, the feature extraction unit 1 in this embodiment
7 will be explained in detail. Reference numeral 41 in the figure denotes a feature extraction area calculation section composed of a normal calculation circuit, which uses the vehicle width information obtained by the vehicle width detection section 16 shown in FIG. Using the plate position information, the scan start and end positions of the feature extraction area within the frame grid 43 are determined based on the feature extraction area relative to the license plate set in advance. 42
1 is an address control section consisting of ordinary adders and comparators, and a counting control section 45 for generating an address for scanning the feature extraction area in the frame memory 43 and storing the extracted features in the storage section 46. Outputs the scanning direction and scanning position. The frame memory 43 is the same as the frame memory 12 in FIG. 1, and stores an image of an automobile as shown in the figure. The area surrounded by a broken line in the figure is an example of a feature extraction area.

Further, numeral 44 denotes a change point detection section which outputs a change point signal when the color changes from white to black or from black to white during scanning, and is constituted by an ordinary register and a comparator.

Furthermore, the counting control section 45 counts the number of change points and consecutive white dots or black dots outputted by the change point detection section 44, and stores them in the storage section 46, and stores the scanning position information from the address control section 42. An address to store each counting result is generated. The storage section 46 is a normal memory in which the characteristics obtained by the counting control section 45 are stored, and the characteristics are checked against the contents of the dictionary memory 19 in the checking section 18 shown in FIG. be.

In the above example, the features to be extracted are the number of transition points from white to black or from black to white on the image, and the distribution of the number of consecutive white or black points, but the invention is not limited to this, and the contour line with the background is used as the feature to be extracted. Structural features may be extracted by, for example, taking a
Identification performance can be further improved by adding vehicle paint color information, axle distance information from a pressure sensor, etc.

〔Effect of the invention〕

As explained above, the vehicle identification device according to the present invention includes an imaging means that optically scans and images the front or rear surface of a vehicle such as a running car, and selects from good images taken by this imaging means. A license plate indexing unit that extracts a vehicle O license plate, a character recognition unit that recognizes characters on the license plate, a feature extraction unit that extracts vehicle characteristics from an image obtained from an imaging means, and this feature extraction unit. Since the present invention includes a collation means for comparing and collating the characteristics of the vehicle obtained from the above with the vehicle type and table type information stored in the storage means, it is effective in significantly improving the accuracy of vehicle identification.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of a vehicle identification device according to the present invention. FIG. 2 is a partial front view showing the front of the vehicle. FIG. 3 is an image diagram scanned by the imaging section in the wheel identification device according to the present invention. FIG. 4 is a block diagram of the feature extraction section in the same embodiment.

Claims (1)

[Claims] an imaging device that optically scans and images the front or rear surface of a vehicle such as a running car; a license plate indexing unit that extracts the license plate of the vehicle from an image captured by the imaging device; a character recognition unit that recognizes characters on a license plate; a feature extraction unit that extracts the characteristics of the vehicle from the image obtained from the image pickup unit; and a storage unit that stores the characteristics of the vehicle obtained from the feature extraction unit. A vehicle identification device comprising a collation means for comparing and collating stored vehicle type and vehicle type information.