JPH0395695A - Number plate detection device - Google Patents

Abstract

PURPOSE:To stabilize the detection of a number plate area even in the case of a few number of digits by discriminating the presence of a miscellaneous picture in an area in which no numeral exists, and verifying a consecutive vehicle number candidate area on the basis of this result. CONSTITUTION:Picture information taken in by a television camera 1 is stored in a picture storage part 5 after being quantized by an A/D converting part 4, and after being binarized, it is sent to a shape measuring part 7 so as to measure shape data, and it is sent to a consecutive vehicle number candidate area detecting part 8 so as to detect the consecutive vehicle number candidate area. A consecutive vehicle number candidate are verifying part 9 sets a detection area for the candidate of a less number of digits among the detected candidate areas, and discriminates whether it is the miscellaneous picture or not according to the size the existing picture, and verifies the consecutive vehicle number candidate area on the basis of this discriminated result.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention (Industrial Field of Application) The present invention provides a method for capturing a scene image including a license plate of a vehicle such as a car with a television camera,
The present invention relates to a license plate detection device that detects a license plate area from the captured scene image.

(Prior art) Recently, for example, on toll roads such as expressways,
2. Description of the Related Art There is a need for technology to automatically read license plate information of passing vehicles (automobiles) for the purpose of preventing illegal passing through the exchange of passes and automatically collecting tolls.

Conventionally, in order to meet such demands, for example, as shown in Japanese Unexamined Patent Application Publication No. 1984-193084, a front image of a passing vehicle detected by a sensor is captured with a television camera, and this captured vehicle image is After the image is binarized using the binarization method, the connected area of the binarized image is analyzed and the license plate area is detected by detecting a serial vehicle number of 4 digits or less in capital letters that is relatively easy to find. There is.

However, with the above-mentioned conventional technology, various miscellaneous images (images other than the number plate) are generated in the binarized image, including the front grill of the vehicle, although it is easier to find than the vehicle classification number. Therefore, there was a problem in that the smaller the number of digits in the serial vehicle number, the more difficult it became to distinguish it from miscellaneous pictures.

(Problems to be Solved by the Invention) As described above, the present invention is capable of detecting a license plate area from a scene image including a license plate. This was developed to solve the problem of difficulty in distinguishing from miscellaneous images, and even when the number of digits in a serial vehicle number is small, the license plate area can be stably detected from binarized images that include miscellaneous images. The object of the present invention is to provide a license plate detection device that can perform the following functions.

[Structure of the Invention] (Means for Solving the Problems) The license plate detection device of the present invention includes a photoelectric conversion means for capturing and photoelectrically converting a scene image including a vehicle license plate, and an output from the photoelectric conversion means. a binarizing means for binarizing image information, a shape measuring means for measuring shape data for each connected region in the image information binarized by the binarizing means, and a shape measuring means for measuring shape data for each connected region; Candidate area detection means for detecting a candidate area for serial vehicle numbers from shape data, and candidate areas with a small number of digits among the candidate areas for serial vehicle numbers detected by this candidate area detection means. Candidate area verification means for determining whether there is a connected area (miscellaneous image) exceeding a predetermined size and verifying the serial vehicle number candidate area detected by the candidate area detection means based on the determination result. are doing.

(Function) Since there is a high possibility that candidates with a small number of digits in the serial vehicle number candidate area are not serial vehicle numbers, all candidates with a small number of digits in the serial vehicle number candidate area, including miscellaneous images in the binary image, are By determining the presence or absence of miscellaneous images in areas where no numbers exist, and verifying serial vehicle number candidate regions based on this determination result, even when the serial vehicle number has a small number of digits,
A license plate area can be stably detected from a scene image including a license plate.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 shows the structure of a license plate detection device according to the present invention, which includes a television camera 1 as photoelectric conversion means, a flash device 2, a vehicle detector 3, and an A/D converter 4.
, image storage section 5, binarization section (binarization means) 6, shape measurement section (shape measurement means) 7, serial vehicle number candidate area detection section (
Candidate area detection means) 8, Series vehicle number candidate area verification unit (
Candidate area verification means) 9, control unit 10 in charge of overall control
It is structured by etc.

That is, first, a scene image of the front or rear side of the vehicle including the license plate of the vehicle is captured by the television camera 1, photoelectrically converted, and captured as image information. At this time,
The image is frozen as a still image with no deviation even when the vehicle is running, for example by using an electronic shutter. Also,
In order to stabilize the lighting conditions, the flash device 2 emits, for example, near-infrared flash light. Note that the timing of capturing image information by the television camera 1 and irradiating flash light by the flash device 2 is as follows.
An incoming vehicle is detected by a vehicle detector 3, and controlled by a control unit 10 based on the detection signal.

The image information captured by the television camera 1 is
After being quantized to, for example, 8 bits by the A/D converter 4, it is stored in the image storage part 5. The image information stored in the image storage section 5 is binarized by the binarization section 6 and then sent to the shape measurement section 7.

Note that, as a method of binarization in the binarization unit 6, for example, a differential image is obtained by subjecting the multivalued image information stored in the image storage unit 5 to spatial quadratic differential processing, and then this differential image is There is a method of binarizing by discrimination using a predetermined threshold value (for example, Japanese Patent Application Laid-Open No. 60-193084).
(see publication).

The shape measurement unit 7 measures shape data such as the coordinates and area of each connected region by performing well-known labeling processing on the image information binarized by the binarization unit 6, and The measured shape data is sent to the serial vehicle number candidate area detection section 8.

The serial vehicle number candidate region detection unit 8 detects connected regions that are candidates for serial vehicle numbers based on the shape data from the shape measurement unit 7 (results of shape measurement of all connected regions), and detects connected regions that are candidates for serial vehicle numbers. A partial image group with a high probability as a serial vehicle number is determined from the positional relationship between them, and this is set as a serial vehicle number candidate area.

For example, as shown in Figure 2, automobile navigation plates generally have characteristics such as the size of each character, the spacing between characters, and how they are aligned in the horizontal direction. By matching, it is possible to detect a serial vehicle number candidate area of the license plate.

Now, the number of characters in the serial vehicle number candidate area detected by the serial vehicle number candidate area detecting unit 8 is generally four digits or less, but the smaller the number of digits, the higher the possibility that it is a false detection of miscellaneous images. Verification as a vehicle number area (license plate area) is required.

In serial vehicle numbers, if the number of digits is small, there are dots (.) in the area where there are no numbers, and conversely, there are no large sketches in this area.

Therefore, the sequential vehicle number candidate area verification unit 9 sets a verification area in the area where dots exist based on the relative positional relationship within the sequential vehicle number candidate area detected by the sequential vehicle number candidate area detection unit 8, and sets a verification area in the area where the dot exists. It is determined whether the image is a miscellaneous image based on the size of the image present in the image, and based on the result of this determination, it is verified that the image is a series of vehicle numbers.

Hereinafter, the processing procedure in the serial vehicle number candidate area verification section 9 will be explained in detail with reference to the flowchart shown in FIG. First, a verification area is set based on the relative positional relationship of serial vehicle number candidate areas (S1). For example, as shown in FIG.
b) As shown in (c), based on the length in the X direction and the length in the Y direction of the serial vehicle number area, the area 11 surrounded by broken lines is determined as the verification area in a method according to the number of digits. do it. In addition, in FIGS. 4(a), 4(b), and 4(c), 12 indicates a vehicle number, and 13 indicates a dot.

Next, in the verification area 11, it is determined whether all connected areas are miscellaneous images (S2).

For example, the dots existing in the verification region 11 set above are very small, and there are not many large images in the verification region 111; , the candidate is determined to be a miscellaneous picture. Specifically, Fig. 4(a)(b)
In the region shown in (c), fixed values are determined for the horizontal width, vertical width, and area of the connected region within the verification region 11, and candidates that do not satisfy the conditions are determined to be miscellaneous images.

Finally, the series vehicle number candidate region that remains without any miscellaneous images in the verification area 11 is determined to be the license plate region (series vehicle number candidate region) (S3). Furthermore, the serial vehicle number candidate area determined to be a miscellaneous image is deleted (S4).

In this way, a serial vehicle number candidate area is detected from the captured image of the front or rear side of the vehicle, including the vehicle license plate, a verification area is set for the candidate with a small number of digits in the detected serial vehicle number candidate area, and this verification area is By determining whether there is a connected region exceeding a predetermined size within the vehicle number and verifying the serial vehicle number candidate region based on the determination result, it is possible to eliminate miscellaneous images even when the number of digits in the serial vehicle number is small. The license plate area can be stably detected from the binarized image.

[Effect of the Invention J As described in detail above, according to the present invention, when detecting a license plate area from a scene image including a license plate, the license plate can be stably detected even when the number of digits in a serial vehicle number is small. A license plate detection device capable of detecting a region can be provided.

[Brief explanation of drawings]

The drawings are for explaining one embodiment of the present invention. Fig. 1 is a block diagram schematically showing the overall configuration, Fig. 2 is a diagram showing an example of a car license plate, and Fig. 3 is a block diagram schematically showing the overall configuration. FIG. 4 is a flowchart showing the processing procedure of the serial vehicle number candidate area verification unit. FIG. 4 is a diagram showing an example of setting the verification area in the serial vehicle number candidate area verification unit. 1... Television camera (photoelectric conversion means),
2...Flash device, 3...Vehicle detector, 4...A/D converter, 5...Image storage unit, 6... ...Binarization unit (binarization means), 7
...Shape measurement section (shape measurement means), 8...
・Serial vehicle number candidate area detection unit (candidate area detection means)
, 9... Series vehicle number candidate area verification unit (candidate area verification means), 10... Control unit. (Note 2) Sat i! (Return setting of (E1)}-Frt, e Eero rate detour n: Mounting 1 piece 5 t member plate Fig. 2

Claims (1)

[Scope of Claims] Photoelectric conversion means for capturing and photoelectrically converting a scene image including a vehicle license plate; Binarization means for converting image information outputted from the photoelectric conversion means into two values; a shape measuring means for measuring shape data for each connected region in the image information binarized by the converting means; and a candidate region detecting means for detecting candidate regions of serial vehicle numbers from the shape data measured by the shape measuring means. , Among the serial vehicle number candidate areas detected by this candidate area detection means, for candidates with a small number of digits, it is determined whether a connected area exceeding a predetermined size exists in the area of digits where no numbers exist, and this determination is performed. A license plate detection device comprising candidate area verification means for verifying the serial vehicle number candidate area detected by the candidate area detection means based on the result.