Interactive Content-Based Medical Image Retrieval System Research And Design

With the development of information technology, driven by the technology of computer vision and the database, the image information has been widely used. There is large number of imaging equipments in the hospitals and research institute. Produce large amounts of image information everyday. The traditional image retrieval is computer-oriented not user-oriented. How to into user’s feedback in the retrieval system is current research focus.In this paper analysis of image texture features and characteristics of mutual information. Use Gabor filter and mutual information extract image’s texture feature and mutual information respectively. Integration of these two features compose a new feature. In the retrieval process results in better effect with the new feature. Improve the retrieval accuracy in the experimental results.Between the low-level visual features and the high-level semantic features there are lots of difference, so the relevance feedback has been introduced to CBIR. And the System learns the requirements of different users from the feedback procedures. Through research the techniques of relevance feedback, the system use the methods of weight adjustment and moving query point. Users according to the system’s retrieval result and submit the feedback information to the system. The system will has the ability to self-learning, and the retrieval result meets the requirements of users step by step.Large amount of information contained in the image, and the image processing is very time-consuming. The retrieval speed is the bottleneck. With the development of computer hardware, Dual-core and multi-core processors have been very common. In order to make better use of system resource, in the retrieval system we add the multi-core parallel computation. The experiment proved that the retrieval speed has been greatly improved.At last, a prototype image retrieval system of medical image retrieval is constructed, which realizes the content-based image retrieval and feedback image retrieval and the multi-core parallel computation.