Image Retrieval by Content using Shape, Color, Texture and Relevance Feedback

Content based image retrieval is a technique used for getting similar type of images to an input image given by a user. Content based image retrieval is used in several areas like medical, biodiversity information system, digital libraries, intellectual properties, crime prevention, security check, remote sensing, engineering and architectural design, publishing and fashion etc. The CBIR manages to provide these applications by using visual features extraction for color, shape and texture features of the images. To provide good quality of results, appropriate features comparison and matching techniques are required. CBIR is the technique of providing similar and accurate results for a query image given by the user. CBIR assumes that all the resulted images on the output screen will be the same as the query image at any instant of time but sometimes it gives false resulted images. So, we can say that the performance of an image retrieval system also depends on the image extraction techniques and the type of images. Most of the existing techniques only consider either the individual features or combined form of features. In this paper, the image retrieval technique uses the best feature extraction technique for the individual feature first and then combined them in a hybrid form and this technique improves the overall precision value of the existing technique by 4%.