【摘要】 针对目前车牌定位和图像提取存在的问题,提出一种在频域中处理的新方法——基于小波伪运动分解的图像分割.首先对车牌图像进行裁剪、灰度化和小波阀值消噪预处理,滤掉一些高频干扰信号;然后,对不同尺度层的小波函数的系数进行小波伪运动分解.最后,根据计算出来的频率特征确定车牌位置,把车牌剪切出来.该算法是基于文字本身的频率来实现的,所以无论背景有多复杂,都可以很好地找出车辆的位置.更多还原

【Abstract】 With the problem of license plate orientation and image extraction,this paper puts forward a new method processed in frequency field,which based on image segmentation of wavelet false movement′ decomposition.At first,we deal with image of license plate,for example,clipping,graying,denoising of wavelet threshold,so that we can filter out high-frequency interference,and then,decompose wavelet false movement with the coefficient of different wavelet function wavelet false movement.At last,we confirm license plate and get it by computing frequency characteristic.This algorithm depends on the frequency of character itself,so no matter how complicated background,it can find out well the position of license plate.更多还原