基于等距环局部二值模式的特征描述子

摘要

Existing descriptors are generally associated with principal direction.When the images are transformed,deviation usually occurs in the principal direction of the descriptors,which leads to unsatisfactory image matching results.In order to solve this problem,a binary descriptor construction method based on Local Binary Pattern (LBP) with equidistant rings is proposed.The sampling model of Equidistant Ring Local Binary Pattern (ErLBP) and matching strategy of grouping are introduced into descriptor constructing and matching,so that the descriptor is rotation-invariant.In this way,the principal direction problem is solved and the speed of descriptor matching is improved.Experimental results show that this method is more accurate and speedy compared with other state-of-the-art binary descriptors,such as BRIEF,BRISK.%现有的描述子一般与主方向相关,当图像经过变换时,得到的描述子主方向会有误差,从而导致图像匹配效果不理想.针对上述问题,基于等距环局部二值模式,提出一种二进制描述子构造方法.在构造描述子与描述子匹配过程中分别使用了等距环局部二值采样模式与分组匹配策略,使得构造的描述子具有旋转不变性,解决主方向问题和加速匹配过程.实验结果表明,与BRIEF,BRISK等二进制描述子相比,该方法具有更快的速度及更高的准确率.