Abstract: This paper proposes a human face shape matching based on shape context useing shape context algorithm. In the shape matching of the two targets similarity comparison, the focus of this approach is to use the shape context algorithm to calculate log-polar coordinate histogram Cost value, the calculation of similarity to the pre-set thresholds to determine whether his face matcheds. Experimental results show that the algorithm in two-dimensional targets in the invariance has accurate matching results

Key words: face match, shape context, log-polar coordinate, Cost value

摘要： 提出一种基于形状上下文的人脸形状匹配算法。在形状匹配中，将2个目标进行相似度比较，得到2幅图的对数极坐标直方图的Cost值，计算相似度是否达到预设的阈值来判定其人脸是否匹配。实验结果表明，该算法在二维目标的不变性中，具有准确的匹配效果。

关键词: 人脸匹配, 形状上下文, 对数极坐标, Cost值

CLC Number: 

• TP18