TY - JOUR AU - C.Callins Christiyana, S.Riyas Parveen, P.Robert, M.Poomani alias Punitha, PY - 2021/05/15 Y2 - 2024/03/28 TI - Impact of Distance Metrics in Local Energy Oriented Pattern Based Image Retrieval JF - Design Engineering JA - DE VL - IS - SE - Articles DO - UR - http://www.thedesignengineering.com/index.php/DE/article/view/1566 SP - 564-573 AB - Algorithms in Content-Based Image Retrieval (CBIR) are being developed based on how the features of the images are represented to retrieve images from image archives. Local Binary pattern based CBIR gains attraction nowadays because of its simplicity and efficiency. But the effectiveness of CBIR also depends on the similarity measure used to retrieve the images. This work aims to analyze the impact of similarity measures concerning distance metrics in the performance of Local Energy Oriented Pattern (LEOP) based CBIR. LEOP is considered for the experiment since LEOP encodes pixel level energy orientations to find minute spatial features of an image, whereas other local patterns rely on only the neighborhood relationship. For each reference pixel in the image, LEOP maps four pixel progression orientations to find the top two maximum energy changes, i.e. two more 3X3 grids are derived from four pixel progressions for each reference pixel. The relationship of pixels derived from three 3 X 3 local patches is used to encode the LEOP pattern. The retrieval efficiency of the LEOP based CBIR system in the ORL image database and the UIUC image database is compared concerning Chi-square and D1 distance measures. The precision and recall measures show that Chi-square based LEOP CBIR system outperforms the D1 based LEOP CBIR system. ER -