【摘要】 用基准图每个像素的三个特征参数即边缘密度、粗糙度、标准差作为评价模板图像匹配的品质指标,结合Fisher线性分类原理把估计区域匹配概率问题转化为对像素的分类问题,提出了一种匹配概率估计的新方法。实验结果表明,该方法可以有效地估计匹配概率,遴选出适合要求的景象匹配区。更多还原

【Abstract】 Three image feature indices (edge density, roughness and standard Dvalue) of every pixel in reference image are taken for the quality index of estimation of template image matching. With Fisher linear classifier, a new approach to estimation of matching probability is proposed which transforms the calculation of matching probability into the classification of pixels. The utility of the method is demonstrated by experiment and suitable matching areas can be sifted out. 更多还原