Content Based Medical Image Retrieval using Relevance Feedback Bayesian Network

Content based image retrieval research has been increased lately due to the needs to maintain high multimedia databases in all computer applications. These retrieves the images from the database based on the feature analysis, such as color, texture and shape. Generally, visual attributes of some feature data are difficult to analysis; it needs a descriptor for organizing and searching the relevant content from it. Scale Invariant Feature and Local Binary Pattern are mostly used efficient visual descriptor helps to analysis the texture, size, shape, color, etc. Relevance Feedback Bayesian network is introduced after SIFT-modified LBP descriptor, to improve the accuracy of the retrieved relevant images with the help of neighbor pixel discrimination from the database. It works on the bases of several key points taken by the Bag-of-Features from the query image and compares it with medical images in the database.