Symmetry-constrained linear sliding co-occurrence LBP for fine-grained leaf image retrieval

Xin Chen, Bin Wang

Abstract: Highlights•A novel local descriptor, named Symmetry-Constrained Linear Sliding Co-occurrence LBP (SCLS-CoLBP), is proposed to depict texture patterns of leaf images.•Itencodes both space symmetry characteristics and directional information of leaf image patterns.•Itis extended to an overall and dense description by incorporating multi-scale, multi-position, multi-orientation and differential information.•The bag-of-words model incorporatedthe proposed smooth-coding schemeis used to aggregate the SCLS-CoLBP local descriptors into image-level representation.•The experiments of applyingthe proposed method to FGLIR tasks consistently show its superior performance over the state-of-the-art methods.