1) Location method based on color information of license plate image [12]. There are four types of license plates in China: white characters on a blue background and yellow background.
Black characters, scarlet letters on white background, and white characters on black background. According to the background information of the license plate, the boundary of the license plate can be accurately located. This method has high recognition filter and strong adaptability, but it is easily disturbed by lighting conditions and background, and the calculation amount is generally large, so it is not suitable for the environment with high real-time requirements.
2) Location method based on edge detection [13]. The change of gray frequency in license plate character region is the most stable feature of license plate region.
Sign, you can use its change to locate the license plate. Firstly, the vehicle image is enhanced, then the edge is extracted, and finally the license plate area is detected by horizontal scanning line and other methods. This method has high positioning accuracy, fast response time, and can effectively remove noise, which is suitable for vehicle images with complex background. However, in the case of serious license plate fading, location will fail because the edge of character strokes cannot be detected.
3) Vehicle location method based on geometric features of license plate [14]. The standard outline size of China license plate is 440* 140, which is rectangular.
The length-width ratio of the whole license plate is about 3: 1. Using this inherent feature, the license plate frame is extracted. This method is only effective when the license plate position is basically horizontal and the border is clear and obvious. However, if the frame of the license plate itself is damaged or incomplete or the collected image deviates from the horizontal angle greatly, it will affect the positioning accuracy, so the application scope is narrow.
4) License plate location method based on spectrum analysis [15]. This method transforms the image from spatial domain to frequency domain for analysis,
Such as DFT transform and wavelet transform. Wavelet analysis can segment images at different resolution levels and perform coarse segmentation at low resolution levels, which saves time and narrows the detection range of fine segmentation. The accurate location of license plate region is realized at high resolution level. However, when there is noise in the vehicle image, it will bring great interference to accurately identify the license plate area and affect the accuracy of license plate location.