Local Shearlet Energy Gammodian Pattern (LSEGP): A Scale Space Binary Shape Descriptor for Texture Classification

Download PDF
Open Webpage

Priya Sen Purkait, Hiranmoy Roy, Debotosh Bhattacharjee

Published: 2019, Last Modified: 09 Nov 2025DoSIER 2019EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: A novel texture feature is proposed in this paper to classify texture images. To represent the local texture feature in different scales and spaces, a novel local gammodian binary pattern (LGBP) is applied on the shearlet transform domain. The main advantage of the proposed gammodian structure is the size of the feature vector. Again, the LGBP is also very effective in capturing local edge information. Finally, the local shearlet energy gammodian pattern (LSEGP) is proposed. The output result of the proposed LSEGP on Outex database shows the effectiveness of the proposed descriptor in texture classification.