Abstract: In order to increase the accuracy and efficiency of image matching for low quality mineshaft video images,a new match algorithm based on gradient space and binary texture is proposed.Firstly，in case of improving accuracy,the feature points and their neighborhood status are quickly examined,the binary texture features and statistical features are combined to construct a rotation-invariant binary texture and oriented gradient histogram (RBT-OGH) descriptor;secondly,according to the classification status of the feature points,and two different matching strategies are used for image matching to improve the working efficiency of the algorithm proposed in this paper.The test results show that the problems of image blurring,illumination variation,low constrast and image deformation existed in the mainshaft video images,therefore,it can be applied to conduct mineshaft video images matching with high efficiency.

Key words: Mainshaft image matching, Feature neighborhood status, Feature descriptor, Similarity measurement