基于TAR的图形拐点识别

对三角形面积表示法(Triangle Area Representation,TAR)描述图形进行讨论,分析TAR-AVG(TAR Average Filtering)和TAR-MED(TAR Median Filtering)的原理.用TAR-AVG衡量图形的外凸与内凹特性,用TAR-MED获得边界曲率信息.结合两种方法的滤波特性,提出TAR-MAMF(TAR Median Average Mixed Filtering)方法,对滤波结果用直方图表示,从直方图输出结果中提取有效的特征参数,确定图形中凹点和凸点的数量以及所在边界的位置,实现图形拐点识别.实验证明,TAR-MAMF方法对图形拐点的定位准确、有效.

A Research of Corner Recognition Based on TAR

A novel triangle area representation(TAR)is proposed in order to depict the boundary of a shape.By analyzing TAR-AVG(TAR Average Filtering) characters of figure's concave and convex can be calculated.By TAR-MED(TAR Median Filtering)curvature information of boundary could be obtained.Combining with the two methods,a new method called TAR-MAMF(TAR Median Average Mixed Filtering)is proposed.The calculation result of TAR-MAMF is analyzed with histogram.According to the parameters retrieved from histogram,the number of concave or convex is determined and each of them is properly located.Meanwhile,the corner recognition of a figure is accomplished.Experiments certify that TAR-MAMF method is accurate and effective.