chrispeng [ad_1]
ITS (Intelligent Transportation System) is a real-time, accurate and efficient transportation management system that uses information and communication technology to closely coordinate and unify the three of people, vehicles, and roads, and establishes a large-scale and all-round role. The system plays an extremely important role in effectively using existing transportation facilities, reducing traffic load and environmental pollution, ensuring traffic safety, improving transportation efficiency, promoting social and economic development, improving people’s quality of life, promoting social informatization, and forming new industries. , Which has been valued by countries all over the world, and has now formed the development direction of world transportation in the 21st century.
As one of the core technologies, license plate recognition (LPR) technology is an important part of many ITS-related application systems such as public security law enforcement systems, highway automatic toll collection systems, urban road monitoring systems, and intelligent parking lot management systems. Greatly simplify human labor, eliminate human interference, reduce or even eliminate the possibility of error. Compared with radio frequency signal recognition and barcode recognition technology, there are two major advantages:
(1) There is no need to install a special barcode or radio frequency identification mark on the car;
(2) The LPR system itself is a recognition system based on video technology, which can easily perform image playback and retrieval.
The automatic license plate recognition system mainly includes working modules such as image acquisition, image processing, license plate segmentation, character recognition, near-end or remote database, and network support.
Connect to computer through video capture card
The use of license plate recognition technology
Urban traffic: traffic statistics at intersections
Intelligent traffic violation monitoring and photo management (electronic police)
Expressway: automatic toll collection, automatic vehicle registration, violation record
Public security system: accident/theft/criminal vehicle monitoring
Military fortresses, institutions, hotels: automated vehicle management
Parking lot and residential area: management of vehicles entering and exiting, automatic billing;
The essential:
Part 2: Location and Segmentation of Vehicle License Plates
That is, the location of the license plate area is found from the image containing the entire vehicle, and the subsequent license plate character recognition is determined.
Part Three: License Plate Character Recognition
Features of license plate images
my country’s existing vehicle license plates: 4 categories-white on blue, black on yellow, white on black, and black on white.
Vehicle license plate features:
(1) A seven-character sequence consisting of Chinese characters for a province (other Chinese characters for military police cards) followed by letters or Arabic numerals. The specific arrangement format of the standard license plate is: X1X2.X3X4X5X6X7,
(2) The color contrast between the bottom of the license plate and the words in the license plate area of the vehicle is large, and the edges are very rich
(3) The height and length of the sub-image area of the vehicle license plate on the image taken at a relatively fixed license plate position are constant, and the proportion of Chang-ho is constant. Original size of license plate: character width 45mm, character height 90mm, spacer width 10mm, and each unit has an interval of 12mm.
Overview of Lipno’s License Plate Location Technology
Starting point: Judge the license plate based on the characteristics of the license plate area.
Main features of license plate:
(1) The edge gray histogram statistics “features” in the license plate area. ——There are two distinct and separated distribution centers.
(2) The geometric characteristics of the license plate, that is, the ratio of width to height of the vehicle is within a certain range.
(3) The gray distribution characteristics of the license plate area, the horizontal straight line passing through the license plate shows a continuous peak, valley, and peak grayscale distribution.
(4) The horizontal or vertical projection characteristics of the license plate area. The horizontal or vertical projection of the license plate area presents a continuous distribution of peaks, valleys, and peaks.
(5) The shape characteristics of the license plate and the character arrangement format characteristics. The license plate has a rectangular frame, and the characters are located in the rectangular frame with spaces.
(6) Spectral characteristics, that is, DFT transformation of rows or columns of graphics. The spectrogram contains the location information of the license plate.
License plate positioning system
A license plate location system usually includes image preprocessing, license plate area search, license plate location and segmentation, etc.
Difficulties of license plate image positioning
(1) The captured image is disturbed by environmental factors (uneven ambient light, etc.), and the photo quality is difficult to guarantee.
(2) Other character areas interfere, making it difficult to locate accurately.
(3) The license plate is stained, dirty, handwriting blurred, faded, etc.
(4) The license plate is partially obscured.
(5) Blurring and distortion of moving images, forming jagged, etc.
Key Points of Lipno Technology’s License Plate Detection and Location Method
1. Grayscale: Convert 24-bit true color images into grayscale images for uniformity and speed with subsequent processing.
2. Gray-scale stretching: The light is insufficient or too strong during imaging, and the image is dark or bright. After processing, the edge of the image is clear, and the stroke characteristics of the license plate area are more obvious.
3. Edge detection: The edge of the license plate area is rich.
4. Template matching: find license plates in images with complex backgrounds
The realization method of Lipno Technology’s license plate positioning
1. Direct method: directly analyze the characteristics of the image
2. Neural network method:
First, the neural network is used to classify the small windows in the image, and then the classification results are integrated to obtain the accurate positioning of the license plate.
3. License plate location method based on vector quantization:
The image is compressed while locating the license plate;
The processing of the image is not in pixels, but in blocks, which improves the processing speed;
It is easy to recognize that there is no license plate in the image.
Recognition of license plate characters
Similar to the general OCR recognition method.
Main algorithm:
1.OCR algorithm based on template matching
First, binarize the character to be recognized and scale its size to the size of the template in the character database, then match it with all the templates, and finally select the best match as the result. Improvement: template matching algorithm based on key points.
5 simple recognizers: simple template matching; peripheral contour matching; improved threading method; template matching based on Hausdorff distance; simple classifier.
2.OCR algorithm based on artificial neural network
Key Points for Realization of License Plate Character Recognition
1. Pretreatment:
(1) Binarization: (color segmentation method)
Difficulty: Threshold selection and license types are diverse (to be unified)
Method: global threshold (OSTU, etc.) and local threshold
(2) Inclination correction: Hough transform detects the inclination of the straight line.
(3) Character segmentation and size normalization: statistical analysis method
2. Character Recognition (OCR)
General matching recognition method, wavelet transform, fractal, etc.
Difficulties of License Plate Character Recognition
The license plate is composed of Chinese characters, letters and numbers. There are many strokes of Chinese characters, the image must have higher resolution, the system must have a high acquisition and processing speed, and it must achieve real-time processing.This requires the adopted algorithm to be concise, practical and high
This article was written by Shenzhen Lipno Technology Co., Ltd. Plagiarism and application by outsiders are strictly prohibited, and offenders will be investigated in accordance with the law!
[ad_2]
