多维彩色图像特征快速抽取方法仿真研究

摘要

为了有效地提升多维彩色图像的识别质量,需要进行多维彩色图像特征快速抽取,但是采用当前方法进行快速抽取时,无法准确的提取图像特征,存在图像抽取误差大的问题.为解决上述问题,提出一种基于小空间占用的多维彩色图像特征快速抽取方法,先采用无迭代双边二维主成分分析方法对多维彩色图像主颜色通道进行颜色特征抽取,将抽取到的各个颜色矩阵融合为一个多维彩色图像的二维矩阵,利用马尔可夫随机场理论抽取一组可以反映图像局部特征的参量,并构造出一组多维彩色图像的特征集,利用快速模糊C-均值聚类方法对图像特征集进行划分,以划分的结果为依据完成对多维彩色图像特征快速抽取.实验结果表明,所提方法抽取精确度高,可以有效地提升多维彩色图像识别效率.%It needs to extract the multi-dimensional color image feature rapidly in order to improve the recognition quality of color image effectively.However,current method cannot extract the image feature accurately during rapid extraction.It leads to big image extraction error.Thus,we propose a rapid extraction method of multi-dimensional color image feature based on the small space occupation.Firstly,the color feature in the main color channel of multi-dimensional color image is extracted by using the non-iterative bilateral two-dimension principal component analysis.Then,each extracted color matrix is fused into a two-dimension matrix of multi-dimensional color image,and a group of parameters reflecting the image local feature are extracted by using the Markov random field theory.A group of feature sets of multi-dimensional color image are built,and the rapid fuzzy c-mean cluster method is used to divide the image feature set.Finally,the rapid extraction of multi-dimensional color image feature is completed according to the division results.The simulation results show that the method has higher extraction precision than the current method.It can improve the recognition efficiency of multi-dimensional color image.