DefinePK

Abstract

The present paper put forward efficient content-based image retrieval (CBIR) system by extracting structural, texture and local features from images. The local features are extracted from local directional pattern (LDP). The LDP produces a steady local edge response in the presence of noise, illumination changes. The LDP coded image is converted in to a ternary pattern image based on a threshold. The structural features are derived by extracting textons on the “local directional ternary pattern (LDTP”) image. The texture features are derived by constructing grey level co-occurrence matrix (GLCM) on the derived texton image. Image retrieval results on various data base images based on various classifiers have proved the discrimination power of the proposed method over existing methods.