Is Hamming distance the only way for matching binary image feature descriptors?
This work addresses a potential improvement in computer vision matching for researchers, but it is incremental as the gains were not statistically significant.
The paper investigated whether metrics other than Hamming distance could improve the accuracy of homography matrices from binary image feature descriptor matching, finding that Jackard-Needham and Dice metrics sometimes performed better but without statistical significance.
Brute force matching of binary image feature descriptors is conventionally performed using the Hamming distance. This paper assesses the use of alternative metrics in order to see whether they can produce feature correspondences that yield more accurate homography matrices. Two statistical tests, namely ANOVA (Analysis of Variance) and McNemar's test were employed for evaluation. Results show that Jackard-Needham and Dice metrics can display better performance for some descriptors. Yet, these performance differences were not found to be statistically significant.