Coupled Bi-Orientation Octet Pattern for Medical Image Retrieval
Yoggya Mehrotra, Shekhar Yadav · 2020
In this paper a novel feature descriptor is proposed named coupled bi-orientation octet (COBOT) pattern for medical image retrieval. COBOT enhances the image with octet encoding so that minute features of an image can be utilized for classification and retrieval. Loss of features is comparatively very less in COBOT when compared with other feature descriptors. COBOT can be explained in terms of orientation and octet. Bi orientation refers to both clockwise and anti-clockwise movements whereas, octet property divides the whole image in eight equal patches and extracts information from each of them. COBOT discern and augment the lines, corner point and texture information in the precinct octet across diverse directions. To further improve the robustness, multiresolution Gaussian filters with COBOT is incorporated and it is termed as MEGABOT. The proposed method transcends various other extensively used hand-crafted feature descriptors used for classification and retrieval of biomedical images.