摘要
针对KMP图像匹配方法应用于带噪声或子图与模板灰度非一致时的图像匹配中存在效率较低,匹配成功率很低的问题,提出了一种基于差分二值矩阵的KMP图像快速匹配算法。该算法先对图像矩阵进行差分求值,利用二值矩阵再进行KMP图像行匹配的方法搜索可能的匹配位置,比较这些位置的整个图像的匹配情况,从中筛选出正确的匹配位置。同时,在行匹配过程中通过记录开始的匹配位置来减少搜索空间,提高效率。实验表明,该方法有效提高了匹配速度,保证了匹配正确率。
Because algorithms based on KMP applying to image matching has low efficiency and low success rate while pictures have noise or grays of template are different from grays of corresponding child picture, algorithm based on two-value subtracting matrix are used to improved it. In this algorithm, numbers in picture matrixes are subtracted using a model matrix firstly, search every row and find the probable matching locations in KMP method with two-value matrix, compare all of these locations and select the most suited location on tile whole. At the same time, marking the preliminary number's location in the process of matching can reduce searching room and improve matching efficiency. Experiment shows that matching velocity is increased without the loss of matching accuracy evidently using this algorithm.
出处
《微计算机信息》
2010年第2期195-197,共3页
Control & Automation
基金
基金申请人:王晓丹
项目名称:SZM集成差异度量及优化技术研究
基金颁发部门:陕西省科技厅(2007F19)
作者简介
林南章(1984-),男,浙江瑞安人.空军工程大学硕士研究生,主要研究方向为图像处理与分析;通讯地址:(713800陕西省三原县空军工程大学导弹学院6系研三队)林南章
王晓丹(1967-),女,陕西汉中人。教授、博士生导师,主要研究方向为模式识别与智能系统,图像处理与分析。