TAILIEUCHUNG - Learning interaction measure with relevance feedback in image retrieval
In this paper, the authors further improve the use of users feedback with multi-feature query and the Choquet integral. Taking into account the interaction among feature sets, feedback information is used to adjust the feature’s relevance weights that are considered as the fuzzy density values in the Choquet integral to define the overall similarity measure between two images. | Journal of Computer Science and Cybernetics, , (2016), 113–131 DOI no. LEARNING INTERACTION MEASURE WITH RELEVANCE FEEDBACK IN IMAGE RETRIEVAL NGO TRUONG GIANG1 , NGO QUOC TAO2 , NGUYEN DUC DUNG2 , NGO HOANG HUY2 1 Faculty 2 Institute of Information Technology, HaiPhong Private University; of Information Technology, Vietnam Academy of Science and Technology; giangnt@,{nqtao,nddung,nhhuy}@ Abstract. Relevance feedback is an effective approach to bridge the gap between low-level feature extraction and high-level semantic concept in content-based image retrieval (CBIR). In this paper, the authors further improve the use of users feedback with multi-feature query and the Choquet integral. Taking into account the interaction among feature sets, feedback information is used to adjust the feature’s relevance weights that are considered as the fuzzy density values in the Choquet integral to define the overall similarity measure between two images. The feature weight adjustment and integration aims at minimizing the difference between users desire and outcome of the retrieval system. Experimental results on several benchmark datasets have shown the effectiveness of the proposed method in improving the quality of CBIR systems. Keywords. Content-based image retrieval, relevance feedback, linear programming, fuzzy measures, Choquet integral. 1. INTRODUCTION Image retrieval has become an important research topic in multimedia applications and attracted attention of many researchers in last decades. Basically, there are two main approaches in building an image retrieval system: text-based and content-based [22, 11, 17]. In text-based image retrieval systems, the users’ queries are composed by key-words, which describe image content. The system retrieves images using image labels which are annotated manually. However, the difficulties in annotating a massive number of images and the avoiding subjectively labeling .
đang nạp các trang xem trước
