Combined ternary patterns for texture recognition
Ni Liu, Georgy Gimel’farb, Patrice Jean Delmas · 2015
Local ordinal signal relations, such as local binary or ternary patterns (LBP/LTPs) are invariant to frequent in practice spatially variant contrast/offset deviations that preserve image appearance. Our prior work extended this conventional LBP/LTP-based classifiers towards learning, rather than pre-scribing characteristic shapes, sizes, and numbers of such patterns. The learned LTPs showed more accurate image query based texture retrieval. But the disadvantage is that the short-range local structures disappeared for some textures. In this paper, we propose to overcome it by combining both the fixed circular LBP/LTP structure and the long-range learned LTPs as texture descriptors and improved performance has been obtained for texture retrieval.