对称二值模式下多层次特征图像匹配算法仿真  

Simulation of Multi-level Feature Image Matching Algorithm in Symmetric Binary Mode

作　　者：张锐敏[1] 周涛[1] ZHANG Rui-min;ZHOU Tao(College of Information Science and Technology,Shihezi University,Shihezi 832003,China)

机构地区：[1]石河子大学信息科学与技术学院

出　　处：《计算机仿真》2019年第12期432-436,共5页Computer Simulation

摘　　要：针对传统方法对多层次特征图像进行匹配时存在抗噪能力差、匹配精度低的问题,提出一种对称二值模式下多层次特征图像匹配算法。通过SIFT特征提取法提取图像特征,利用尺度集对图像特征的极值进行筛选,并得到有效特征点。以纹理特征、颜色特征和方向特征为指标对有效特征点进行描述,使图像特征更具清晰化。根据描述结果建立k-d树,对多层次图像特征粗匹配,进而完成多层次特征图像匹配。为检测匹配算法效果,与传统匹配算法进行对比,仿真结果表明,所提方法能够在短时间内较为精准地完成匹配,抗噪能力强,匹配效果好。Due to low anti-noise ability and low matching precision in the traditional methods,this article puts forward a multilevel feature image matching algorithm based on symmetric binary mode.At first,the image features were extracted by SIFT feature extraction method.Then,extreme values of image features were screened by the scaling set,and the effective feature points were obtained.Moreover,the effective feature points were described by texture feature indexes,color feature indexes and direction feature indexes,so that the image features would be clearer.According to the description result,k-tree was established to match the features of multilevel image in rough,and thus to complete the multilevel feature image matching.In order to detect the effect of the matching algorithm,the proposed algorithm was compared with the traditional matching algorithm.Simulation results show that the proposed method can complete the matching more accurately in a short time.Meanwhile,the method has strong anti-noise ability and good matching effect.

关 键 词：对称二值模式 多层次特征 图像匹配 匹配算法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]