Abstract

In this abstract we extend ideas and results submit-ted to [3] in which a new codification of Local Binary Patterns(LBP) is given using combinatorial maps and a method forobtaining a representative LBP image is developed based onmerging regions and Minimum Contrast Algorithm. The LBPcode characterizes the topological category (max, min, slope,saddle) of the 2D gray level landscape around the center region.We extend the result studying how to merge non-singular slopeswith one of its neighbors and how to extend the results to non-well formed images/maps. Some ideas related to robust LBP andisolines are also given in last section.

Reference

González-Díaz, R., Kropatsch, W., Cerman, M., & Lamar, J. (2014). Characterizing configurations of critical points through lbp. In D. M. Onchis & P. Real Jurado (Eds.), Proceedings of the 5th International Workshop on Computational Topology in Image Context 2014 (p. 4). Proceedings of the 5th International Workshop on Computational Topology in Image Context 2014. http://hdl.handle.net/20.500.12708/55280