This paper presents a novel binary descriptor named as B-SIFT(Binarized Scale Invariant Feature Transform) for efficient invariant feature correspondence.Through analyzing the local distinctive gradient pattern, we convert the standard SIFT descriptor to a binary representation which can be computed extremely fast with bitwise operation. Extensive correspondence trials based on a benchmark Oxford image data set with viewpoint, scale, image blur, JPEG compression and illumination changes demonstrate that in general, the proposed B-SIFT method significantly outperforms the standard SIFT with over 400 times faster in matching time and 32 times less in memory resources, while achieves the same matching score as SIFT. © 2012 Springer-Verlag.
CITATION STYLE
Li, J., & Lu, Z. (2012). B-SIFT: A highly efficient binary SIFT descriptor for invariant feature correspondence. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7202 LNCS, pp. 426–433). https://doi.org/10.1007/978-3-642-31919-8_55
Mendeley helps you to discover research relevant for your work.