KR20100005421A - Image-processing-based automatic vehicle identification system of non-detection type - Google Patents

Abstract

PURPOSE: An image-processing-based automatic vehicle identification system of a non-detection type are provided to reduce installation costs by detecting a vehicle trough image processing of a photographed image and make system installation simple. CONSTITUTION: A camera image input unit(10) receives images of at least one lane which is photographed by a camera. A vehicle area detector(20) performs a digital image process of an image from the camera image input unit and extracts a vehicle from the image. A number plate area detector(30) detects a number plate area of vehicles from the vehicle image data. A character/number area detector(40) extracts a character and number region from the number-plate image data of the vehicles.

Description

Image-processing-based automatic vehicle identification system of non-detection type}

The present invention relates to a vehicle recognition system. More particularly, the present invention relates to an image-based non-detection automatic vehicle recognition system capable of recognizing a vehicle by signal processing a captured image using non-point knowledge, rather than a conventional detection method.

Automatic Vehicle Identification (AVI) refers to a technology for automatically recognizing a vehicle license plate. The vehicle license plate information obtained through this process is processed as basic data of an intelligent transport system (ITS). This automatic vehicle recognition technology is a basic step that must be performed to obtain necessary traffic information from an image obtained from a camera installed on a road.

Conventional AVI systems are detectable, which refers to a system equipped with a module that continuously detects whether the vehicle is in or out. Then, when the entrance and exit of the vehicle is detected, there is a module that receives a signal from the detection module and takes a picture. The captured image is transferred to a recognition module that recognizes the license plate to recognize the license plate of the vehicle. The recognized license plate data is used as basic information for later use in intelligent transportation networks. There are various types of such detection systems according to the method of detecting the entry and exit of the vehicle, which is divided into a buried type and a non-buried type based on whether the detection device is embedded in the lane.

FIG. 1 illustrates a buried system in which a loop coil is embedded in a lane as a detection device in a conventional detection type AVI system, and FIG. 2 illustrates a laser beam as a detection device instead of embedding a loop coil in a lane in a conventional detection type AVI system. A non-embedded system using generation and inspection equipment is shown.

Such a detection AVI system has the following problems.

First, the system installation and migration is complicated. That is, in the case of buried system, when installing the system, complicated road construction is required due to the embedding of the detection loop coil in the roadway, and it is complicated by moving the roof coil and reburying the roof coil even when the system is transferred. Road work is needed. In addition, even in the case of the non-embedded system, the installation of the laser beam generation and inspection equipment as well as the installation of the camera must be installed at the time of installation of the system.

Secondly, installation costs are high. That is, in the case of the buried type, expensive loop coils and connecting lines are required, and expensive road construction costs are incurred. In the case of the non-embedded type, expensive laser beam generation and inspection equipment is required.

Third, it is difficult to expand the lane to which the license plate is recognized. In other words, in order to expand the lane to be recognized by the license plate, installation and operation costs increase because detection and photographing modules need to be added for each lane.

Fourthly, in the case of the laser beam method, the license plate cannot be recognized from multiple lanes by installing on the side of the road.

The present invention has been made to solve the above-mentioned conventional problems, the installation and transfer of the system is not complicated and the installation cost is relatively inexpensive, and can be installed on the side of the road without any additional cost even when expanding the target car to recognize An object of the present invention is to provide an image-based non-detection automatic vehicle recognition system capable of recognizing a vehicle of multiple lanes.

In order to achieve the above object, an image-based non-detection automatic vehicle recognition system according to the present invention includes a camera image input unit for receiving an image of at least one lane photographed by a camera and outputting the image as digital image data; A vehicle region detector for extracting a vehicle image by detecting a vehicle through digital image processing of the digital image data input from the camera image input unit; A license plate area detector for detecting a license plate area of a vehicle from the vehicle image data output from the vehicle area detector; A character / numeric region detector for extracting a character and numeric region from the license plate image data of the vehicle output from the license plate region detector; And a character / number recognition unit for recognizing characters and numbers from the character and numeric image data output from the character / number region detection unit and outputting them as vehicle number data.

Here, the image-based non-detection type automatic vehicle recognition system according to the present invention may further include a vehicle specific information detector for detecting vehicle specific information from the vehicle image data output from the vehicle region detector. The vehicle specific information detection unit may further include: a vehicle size measuring unit configured to measure the size of the vehicle from the vehicle image data output from the vehicle region detection unit and output the measured size as vehicle size information; A vehicle color recognizer configured to recognize a color of the vehicle from the vehicle image data output from the vehicle region detector and to output the recognized color as vehicle color information; And at least one of a vehicle headlight detection and interval measuring unit configured to detect the headlights of the vehicle from the vehicle image data output from the vehicle region detection unit, measure the distance between the detected left and right headlights, and output the headlight interval data. Can be.

The vehicle region detector may include: a time delay unit configured to delay and output digital image data output from the camera image input unit for a predetermined time; A foreground and background separator for separating and removing the foreground and the background by comparing the digital image data output from the current camera image input unit with the delayed digital image data output from the time delay unit; A vehicle motion estimation and compensation unit which receives digital image data from which the foreground and background are separated from the foreground and background separation unit, calculates a motion vector by using a motion estimation and compensation algorithm, and estimates and compensates the movement of the vehicle; And a vehicle detector configured to receive image data compensated for the movement of the vehicle from the vehicle movement estimator and compensator to detect an optimized vehicle region.

According to the present invention configured as described above, compared with the conventional method, the installation and transfer operation of the system is not complicated and the installation cost is relatively low, and it is possible to cope without incurring additional costs even when expanding the lane to be recognized, and installed on the side of the road. It is possible to recognize a vehicle of multiple lanes.

Hereinafter, an image-based non-detection automatic vehicle recognition system according to a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings.

3 is a view for explaining an image-based non-detection automatic vehicle recognition system according to the present invention.

As shown in the drawing, the image-based non-detection automatic vehicle recognition system according to the present invention, unlike the conventional detection method, does not detect a vehicle using a loop coil or a laser beam, but images captured by a camera. Based on the data, image-based detection technology is used to detect vehicles passing through multiple lanes.

4 is a block diagram illustrating an image-based non-detection automatic vehicle recognition system according to the present invention, and FIG. 5 is a block diagram illustrating a detailed block configuration of the vehicle region detection unit of FIG. 4.

As shown in FIG. 4, the image-based non-detection type automatic vehicle recognition system according to the present invention includes a camera image input unit 10, a vehicle area detection unit 20, a license plate area detection unit 30, and an alphanumeric area detection unit ( 40), an alphanumeric recognition unit 50 and a vehicle specific information detection unit 60 are configured.

The camera image input unit 10 receives an image of a plurality of lanes photographed by the camera and outputs the digital image data in frame units.

The vehicle region detector 20 detects a vehicle through digital image processing of frame-based digital image data input from the camera image input unit 10 and extracts a vehicle image.

As shown in FIG. 5, the vehicle area detector 20 includes a time delay unit 24, a foreground and background separator 22, a vehicle motion estimation and compensator 26, and a vehicle detector 28. It may be provided with. The time delay unit 24 delays and outputs the digital image data of the frame unit output from the camera image input unit 10 for a predetermined time (for example, one or two frames in frame unit time). The foreground and background separator 22 compares a current frame of the digital image data output from the current camera image input unit 10 with a previous frame of the delayed digital image data output from the time delay unit 24 to compare the foreground and the background. The digital image data is separated and removed, and the foreground and background are separated and output. The vehicle motion estimation and compensator 26 receives digital image data from which the foreground and background are separated from the foreground and background separator 22, calculates a motion vector using a motion estimation and compensation algorithm, and calculates the motion of the vehicle. Estimate and compensate. The vehicle detector 28 receives the image data compensated for the movement of the vehicle from the vehicle motion estimation and compensator 26, detects an optimized vehicle region, and outputs only the detected image data of the vehicle.

The license plate area detector 30 detects the license plate area of the vehicle from the vehicle image data output from the vehicle area detector 20 and outputs only the detected license plate image data of the vehicle.

The character / numeric region detector 40 extracts a character and numeric region from the license plate image data of the vehicle output from the license plate region detector 30 and outputs only the extracted character and numeric image data.

The letter / number recognition unit 50 recognizes letters and numbers from the letter and number image data output from the letter / number area detection unit 40 and outputs them as vehicle number data.

The vehicle specific information detection unit 60 detects and outputs vehicle specific information from the vehicle image data output from the vehicle region detection unit 20. In the present embodiment, only the size of the vehicle, the color of the vehicle, and the interval between the vehicle headlights are described as the vehicle-specific information, but other vehicle-specific information is also possible.

The vehicle specific information detecting unit 60 includes a vehicle size measuring unit 62, a vehicle color recognition unit 64, and a vehicle headlight detecting and spacing measuring unit (headlight area detecting unit 66 and headlight spacing measuring unit 68). can do.

The vehicle size measuring unit 62 measures the size of the vehicle (for example, vehicle length and width) from the vehicle image data output from the vehicle region detecting unit 20 and outputs the measured size as vehicle size data.

The vehicle color recognizer 64 recognizes the color of the vehicle from the vehicle image data output from the vehicle region detector 20 and outputs the recognized color as the vehicle color data.

The headlight area detection unit 66 detects the headlight of the vehicle from the vehicle image data output from the vehicle area detection unit 20, and the headlight spacing measuring unit 68 detects the headlight area of the vehicle detected by the headlight area detection unit 66. The headlight spacing (left and right headlight spacing) is measured and output as headlight spacing data.

As described above, according to the present invention, by overcoming the problems of the conventional detection method by using an image-based vehicle detection technology which is a non-detection method, the multi-lane vehicles are detected and the unique information (color, size, The use of the left and right headlight spacing, etc., in addition to the license plate recognition greatly improves the reliability of the license plate recognition. In addition, according to the present invention, a continuous camera image is acquired in real time, the previous image is compared with the current image, the foreground and the background are separated, the motion vector of the vehicle is calculated, and the optimized vehicle region is detected. When detecting, the license plate image is searched within the minimum vehicle area including the license plate, which greatly improves the processing speed and significantly reduces the false recognition rate.

On the other hand, the present invention is not limited to the above specific embodiments, but can be modified and modified in various ways without departing from the gist of the present invention. If such changes and modifications are included in the appended claims, of course, they belong to the present invention.

FIG. 1 is a view showing a buried system in which a loop coil is embedded in a lane as a detection device among conventional detection automatic vehicle recognition (AVI) systems.

FIG. 2 is a diagram illustrating a non-embedded system using a laser beam generation and inspection apparatus as a detection apparatus, rather than embedding a loop coil in a lane among conventional AVI systems.

3 is a view for explaining an image-based non-detection automatic vehicle recognition system according to the present invention.

4 is a block diagram of an image-based non-detection automatic vehicle recognition system according to the present invention.

FIG. 5 is a diagram illustrating a detailed block configuration of the vehicle region detector of FIG. 4.

<Explanation of symbols for the main parts of the drawings>

10: camera image input unit 20: vehicle area detection unit

22: foreground and background separator 24: time delay

26: vehicle motion estimation and compensation unit 28: vehicle detection and tracking unit

30: license plate area detection unit 40: alphanumeric area detection unit

50: character / number recognition unit 60: vehicle specific information detection unit

62: vehicle size measurement unit 64: vehicle color recognition unit

66: headlight area detection unit 68: headlight spacing measuring unit

Claims (4)

A camera image input unit which receives an image of at least one lane photographed by a camera and outputs the image as digital image data; A vehicle region detector for extracting a vehicle image by detecting a vehicle through digital image processing of the digital image data input from the camera image input unit; A license plate area detector for detecting a license plate area of a vehicle from the vehicle image data output from the vehicle area detector; A character / numeric region detector for extracting a character and numeric region from the license plate image data of the vehicle output from the license plate region detector; And a character / number recognition unit configured to recognize letters and numbers from the character and numeric image data output from the character / numerical area detection unit and output them as vehicle number data. The method of claim 1, And a vehicle specific information detection unit configured to detect vehicle specific information from the vehicle image data output from the vehicle region detection unit. The method of claim 2, The vehicle specific information detection unit, A vehicle size measuring unit measuring the size of the vehicle from the vehicle image data output from the vehicle region detecting unit and outputting the measured size as vehicle size information; A vehicle color recognizer configured to recognize a color of the vehicle from the vehicle image data output from the vehicle region detector and to output the recognized color as vehicle color information; And at least one of a vehicle headlight detection and interval measuring unit configured to detect the headlights of the vehicle from the vehicle image data output from the vehicle region detection unit, and measure the distance between the detected left and right headlights and output them as headlight interval data. Image-based non-detection automatic vehicle recognition system, characterized in that. The method according to any one of claims 1 to 3, The vehicle area detection unit,  A time delay unit for delaying and outputting digital image data output from the camera image input unit for a predetermined time; A foreground and background separator for separating and removing the foreground and the background by comparing the digital image data output from the current camera image input unit with the delayed digital image data output from the time delay unit; A vehicle motion estimation and compensation unit which receives digital image data from which the foreground and background are separated from the foreground and background separation unit, calculates a motion vector by using a motion estimation and compensation algorithm, and estimates and compensates the movement of the vehicle; And a vehicle detection unit configured to detect an optimized vehicle area by receiving image data compensated for the movement of the vehicle from the vehicle motion estimation and compensation unit.

Images