Batik Image Retrieval Using Maximum Run Length of Completed Local Binary Pattern and Symbiotic Organisms Search

This paper presents a new method for performing Batik image retrieval system. The proposed method employs the Completed Local Binary Pattern (CLBP) and Maximum Run Length (MRL) in the similarity matching. The MRL computes the maximum number of repeated occurrence state ‘0’ or ‘1’ in the LBP binary string. Fusion of MRL and CLBP, namely MRL-CLBP, reduces the dimensionality of image descriptor compared to the original CLBP feature. Noteworthy, the performance accuracy of the proposed MRL-CLBP feature is almost the same as the original CLBP descriptor. The proposed system also involves the color features in MRL-CLBP to further improve the accuracy of the content-based Batik image search. The optimization process on the similarity weight constants using the Symbiotic Organisms Search (SOS) algorithm is also able to increase the image retrieval performance. The Experimental Section reports the effectiveness of the proposed image feature with lower dimensionality compared to the former existing scheme in terms of average precision and recall rates. In addition, the MRL-CLBP of state ‘0’ and ‘1’ yields almost similar retrieval performance.