基于整体特征的鞋印匹配算法

根据鞋印图像自身的特点,提出了一种新的鞋印定位方法;在此基础上提出了将鞋印的外纹轮廓简化成四边形,然后用四边形原理的方法来提取鞋印的几何特征用于鞋印匹配;内部纹理特征的识别采用改进的局部二值模型(local binary patter,LBP),为了充分的利用鞋印上的特征;将这内外特征相结合形成一个整体特征用于鞋印的匹配。实验结果表明,整体特征在保持算法鲁棒性情况下,匹配用时短、识别率高。

The Overall Characteristics of Shoe Matching Algorithm

According to the characteristics of the shoes print image itself, this paper proposes a new shoes locating method. On the basis of this, it is put forward that the outer contour of the shoe print is reduced to a quadrangle, and then the geometric feature of the shoe print is extracted by the quadrangle method for shoe print matching; The internal texture feature recognition with the improved Local Binary Patter; In order to make full use of the characteristics of the shoes on, in this paper, the internal and external features combined to form a whole match for shoes print; Experimental results show that the overall features in keeping the algorithm robustness, matching short, high recognition rate.