基于小波变换的字符图像分割

Image Segmentation of License-Plate Based on Wavelet Transformation

为了消除明暗不均等低频干扰的影响,将字符成功地从车牌图像中提取出来,提出了基于小波分解的车牌图像分割方法,根据干扰信号与字符信号属于不同频域的特点,利用小波的时频局部化分析能力,在保留字符信息的同时滤除或抑制低频干扰成分。首先采用逐行逐列扫描的方法对图像预处理后的图像进行车牌定位,再利用基于小波变换的阈值分割法对车牌进行字符分割。

In order to get rid of the low frequency distribution like light and shade and obtain characters from license-plate images, a method is put forward to segment license-plate images based on wavelet decomposes in this paper. Because of the diffident frequency of license-plate characters and disturbing signal, low frequency disturbing signal can be eliminated or rest rained by wavelet transformation, first of all, the area of the license-plate is oriented by the ways of scanning the row and the line of images after pretreatment, and then the character of license-plate is divided up using threshold-division based on wavelet transformation.