Interactive image retrieval system based on multiple features
Yichen Xie · 2016
The most commonly image retrieval method is based on the text at present. The big workload is one of the disadvantages of this approach, it also needs to add tags artificiality, which has a strong subjectivity. More and more people developed themselves into the research of content-based image retrieval technology, and so far there is not a universal retrieval method for any image. In this paper, a interactive retrieval method is realized, and the thought of the optimizing the feedback results by the human-computer interaction is put forward. Some methods for eigenvalue extraction such as the edge detection, the extraction of gray histogram and brightness histogram are studied. These algorithms are also used in the image retrieval system. Experimental results on the STANFORD VERSION LAB 101 Object Categories image library, show the efficiency of the proposed methods. The structure of the paper is as follows: chapter one is the introduction, chapter two is about the five feature extraction algorithms used in the system, chapter three is the idea of eigenvalue storage, chapter four is the experiment on this system, and chapter five is the conclusion.