MBR-SIFT : A Novel Descriptor for Image Matching

The traditional SIFT method is capable of extracting distinctive feature for image matching. However, it is extremely time consuming in the SIFT matching due to applying the Euclidean distance measure. Recently, many binary SIFT (BSIFT) methods have been developed to improve matching efficiency, while none of them is invariant to mirror reflection. To address these issues, this paper presents a mirror reflection invariant binary descriptor, named MBR-SIFT. In addition, the Hamming distance has been employed as similarity measure for MBR-SIFT descriptor. The method is compared with other BSIFT methods, and is shown that MBR-SIFT is better both in accuracy and efficiency.