An effective texture descriptor for retrieval of biomedical and face images based on co-occurrence of similar center-symmetric local binary edges

In this paper, a new texture descriptor is proposed for face and biomedical image retrieval. The proposed texture descriptor extracts the co-occurrence of similar center-symmetric local binary edges which are computed at radius ‘1’ and ‘2’ of the local neighborhoods. In contrast, one recent related work first extracts the local pattern of the original image using center-symmetric local binary pattern (CSLBP) and then applies gray-level co-occurrence matrix (GLCM) with four directions as a result of which four matrices are obtained in each direction. The proposed descriptor with much less dimensions captures the co-occurrence information between the local CSLBP edges very efficiently and shows encouraging discriminativeness over similar techniques. Finally more spatial information is incorporated by forming the feature vector by concatenating the histograms calculated from non-overlapping blocks of the pattern map. The efficacy of the proposed descriptor is tested under face and biomedical image retrieval in terms of precision and recall. The benchmark databases ORL face, OSIRIX-CT, NEMA-CT, and York MRI are used for performance evaluation and comparison of the proposed technique with other similar techniques. © 2019 Informa UK Limited, trading as Taylor & Francis Group.