基于轮廓特征与文字特征的二维碎片匹配

2D Fragments Matching Based on Contour Features and Character Features

针对二维不规则实物碎片,提出了一种基于多种碎片特征的碎片匹配算法.对碎片进行预处理,提取碎片的轮廓特征以及文字特征.文中主要采用高斯滤波对图像进行平滑,基于数学形态学边缘检测算法进行边缘检测,基于曲率尺度空间进行角点检测.通过提取角点处的灰度值进行刻画碎片的文字特征.对提取的角度特征和灰度特征进行加权,应用模式匹配算法寻找其公共子列进行特征匹配.通过数值实验,可以得到较好的匹配效果,且算法简单,降低了匹配误差.

An algorithm for fragments matching based on contour features and character features is proposed to solve the problem of 2D irregular fragments. The fragments are preprocessed first,and then their contour features and character features are extracted. In this study, Oaussian filter was used for image smoothing,the mathematical morphology edge detection for the edge detection,and the curvature scale space for the corner detection. Grey values of corners were extracted to describe the character features. Then the contour features and character features were weighted. Finally the pattern matching algorithm was used to find the public column to match the features. The results show that the proposed algorithm is simple and reduces matching errors with good matching result.