Best-Buddies Similarity Algorithm (BBS) proposed by [1] outperforms commonly used methods for template matching
such as normalized cross correlation, histogram matching and EMD. Several of its key features makes it robust to outliers
and complex geometric deformations. Nevertheless, while original BBS takes local maximums into consideration to find
the best candidate, we introduce an area-based decision driven method to locate a template match in a target image. Our
method shows better success rate than the original BBS on a challenging real-world dataset while managing computational
time to be only marginally increased.
such as normalized cross correlation, histogram matching and EMD. Several of its key features makes it robust to outliers
and complex geometric deformations. Nevertheless, while original BBS takes local maximums into consideration to find
the best candidate, we introduce an area-based decision driven method to locate a template match in a target image. Our
method shows better success rate than the original BBS on a challenging real-world dataset while managing computational
time to be only marginally increased.