A Multi-query Strategy forContent-based Image Retrieval

Content-Based Image Retrieval (CBIR) leads a central role in the multimedia database. The traditional CBIR is often established on single-query image method. However, multi-query has been applied to capture information more precisely in CBIR. There are many applications are interesting on the multi-query. For example, a registered 3D-trademark has different angles images for recognition. A stereo image is constituted by two images from different viewpoint. A set of images helps to represent the scene or trademark. Multi-query strategy will be a suitable mechanism for the retrieval relevant 3D-trademark or stereo image in CBIR. Each image in multi-query has its visual representation in user's respect and major low-level features. How to integrate multi-query properly in the query processing is an important issue. This study proposes Multi-Query Image Retrieval (MQIR) to support multi-query and low-level feature integration. The relevant images are selected based on user's query preference in multi-query. Experimental results present that the proposed approach is more accurate to improve retrieval performance and provides better retrieval results. The proposed multi-query approach was provided to help the query processing in CBIR system.