Feature Based Image Retrieval Using Fused Sift and Surf Features

Abstract

In Computer Vision, the content based image retrieval (CBIR) is an emerging technique applied for image retrieval and also used in most of the search engines. In any feature based image retrieval methods, the extracted features and feature vectors are the essential key for image retrieval. The scale invariant feature transform (SIFT) and the speeded up robust features (SURF) are well-proven methods to extract features from images and those features can also be applied for image retrieval. But both of these methods have their own advantages and disadvantages when applied to image retrieval. By fusing the features extracted by SIFT and SURF, the proposed method forms a new feature vectors by computing histograms for the same and pass it to support vector machine (SVM) classifier for training. The K-means and k-d tree algorithms are also used for optimizing the retrieval results. The results of the parent methods and the proposed method are tabulated and compared by keeping the affecting parameters as same and also taking into account the anomalies. The proposed method is seen to give the best retrieval results in all the cases.