With the improvement of license plate recognition technology, the automatic license plate recognition system is derived, which can automatically recognize the license plate information from the moving vehicle. So what’s the complicated process of license plate recognition.
To recognize the license plate clearly, it is necessary to analyze the image containing the license plate based on the image segmentation and image recognition theory. And then can determine the position of the license plate in the image. After that, further to extract and recognize the text characters. It can be seen that the recognition process of automatic number plate recognition system includes a series of algorithms such as image acquisition, pre-processing, license plate positioning, character segmentation, character recognition and result output.
The first is image acquisition of the automatic license plate recognition system. It’s through the high-definition camera shooting license plate. The picture quality is the key factor affecting the license plate recognition rate. So, it’s necessary for the original image noise filtering, automatic white balance, automatic exposure and gamma correction, Edge enhancement, contrast adjustment and other pre-treatment.
Next is the license plate positioning of ANPR system, its accuracy or not determine the effect of the subsequent character segmentation and recognition. It’s the important factor as the entire car number plate recognition rate. Performing a row scan on a gray scale image after image pre-processing. And then the region of the column scan to determine the column coordinates and width. Finally, a license plate area can be determined.
After the license plate area is positioned in the image, the automatic license plate recognition system will enter the next session - character segmentation. According to the character size feature, the dynamic template method is proposed for character segmentation, the license plate recognition and the character size is normalized by gray scale, gray scale stretching, binarization and marginalization.
Then it is the scaling of the characters after division, feature extraction, access to specific characters in the form of expression. After that through the classification of discriminant function and classification rules, and character database template in the form of standard character expression matching to identify, you can identify the input characters image.
This is the entire work flow of the ANPR system. Through this series of processes, you can accurately vehicle number plate recognition. Intelligent traffic, highway bayonet, parking management, electronic police and so can play an important role in rely on such a software.
If you are interested in our products, please contact us!
Wiicontrol Information Technology Co. , Ltd
Add:Yangming Industrial Park, Liangxi District, Wuxi City, Jiangsu Province, China.