基才奇异值分解的车牌特征提取方法研究与实现

摘要

This thesis studies and analyses the methods of images location, division and size normalization based on wavelet analysis. It also analyses the mathematical principle and algorithm of SVD algorithm.Using SVD as the method of algebraic features extraction, get the useful features description of the Licence Plate images. Extract and obtain the singular value (SV) feature of licence plate character in the Matlab development environment.Calculation has stability and is to the benefit of making the result of SV feature visible. Meanwhile it is convenient to analyse SV feature of licence images.This feature can express the details and structure characteristics of the images. Through the datas and Curve analysis, it can prove that licence character feature has better separability.%研究了基于小波分析的车牌图像定位、分割、大小归一化方法。并分析了奇异值分解算法的数学原理和算法。利用奇异值分解作为代数特征提取方法，获得图像的有效特征描述。以Matlab为开发环境，进行车牌字符的奇异值特征提取，得到车牌字符的奇异值特征，计算具有稳定性，并且有利于将得到的奇异值特征结果可视化，便于对车牌图像的奇异值特征进行分析。该特征能够较好地表达字符图像的细节和结构特征，通过数据和曲线分析，证明车牌字符特征具有较好的可分性。