Local Binary Pattern and Tchebychev Moments as a Single Image Discriptor
Lamine Benrais, Nadia Baha · 2018
This paper experiences a novel approach of object description using a combination of two well-known descriptors "LBP: Local Binary Patterns" and "DTMs: Discrete Tchebychev Moments". With their pros and cons, the LBP and Tchebychev are widely and successfully used in the computer vision community. LBP is a local based descriptor while the DTMs is a global based descriptor seem to makes the combination difficult, however, we managed in this paper to propose an interesting approach that takes advantage of their pros while imitating their cons and outperform their weaknesses. The proposed approach is tested on the COIL dataset and return very interesting results going up to 89,98% of well classified objects.