Based On Significant Weight And Corner Image Retrieval

With the advancement of technology, the popularity of image capture devices such as digital cameras and cell phones, the image can be obtained quickly and easily. At the same time, with the rapid development of Internet, more images are uploaded to the Internet quickly, the image database is expanded at an alarming rate. How to find the picture people need from the image database getting large every day is an important issue. In recent years, content-based image retrieval has been one of the hotspots studied by scholars at home and abroad.In this paper, lots of research work has been done around image retrieval based on saliency weighted and features of corner points.This paper first introduces the critical technology of CBIR systematically, including feature extraction and similarity measure methods, focusing on making a detailed presentation of several commonly used feature extraction methods of color and texture and shape features, for similarity measure, mainly introduces several commonly used methods for vector distance calculating.Secondly, based on the features of corner points of an image, we propose a novel method for image retrieval based on features of corner points. This method makes a special effort to study the corner points of an image, combining texture feature and shape feature as the feature of an image. The feature extraction of the method contains three steps:Harris corner extraction, texture feature extraction of corner points and shape feature extraction of corner points. The weighted Euclidean distance method is used for similarity measurement. Experimental results show that this method is indeed more effective in image retrieval.Finally, saliency feature is added to the method for image retrieval based on features of corner points, the method for image retrieval based on saliency weighted and features of corner points is proposed. This method uses the saliency algorithm to calculate the saliency map corresponding to original image. In the saliency map, the value of each point represents the saliency value of the point in original image. According to this characteristic, the feature extraction is improved. The texture features and the shape feature are weighted by saliency map in this method. By this, the useful information is reserved, and the useless information is inhibited. Experimental results show that this method can form a better description of the image, the image retrieval efficiency has also been improved.