基于概率密度分割的特征约束角点匹配方法

为克服Harris算子特征点匹配的角点群聚现象, 提出了一种基于概率密度的角点匹配算法。该方法将角点间的图像距离作为基本区域划分的主要参考系数, 利用划分区域的角点概率密度减少匹配区域, 然后将区域外的特征点判定为伪角点并将其去除。实验表明, 该改进算法的匹配结果有效地减少了干扰点, 从而提高了算法的实时性和准确性。

Characteristics Constrained Corner Matching Method Based on Probability Density Segmentation

To overcome comer clustering phenomenon of Harris operator feature point matching, we proposed a probability density-based comer matching algorithm. This method sets the image distance between the comer as a basic reference coefficient of the main regional division, using the comer probability density of regional division to reduce the matching area, and judging the feature points outside the region as false comers and removing them. Experiments show that the matching results of improved algorithm effectively reduce interference points, improving the timeliness and accuracy of the algorithm.