Improved SIFT features in image retrieval using

In this paper, based on SIFT feature algorithm, the SIFT feature descriptor is improved, the area of feature points used to construct a new circular region feature descriptor, thus improving the characteristics of the problem of high dimension, but also enhanced features descriptor of their own anti-rotation, and then two images by calculating the feature vector Euclidean distance, the findings were compared in order to achieve image matching, in building the image database to find the minimum distance vector with which to achieve the target image retrieval. Experimental results show that the algorithm not only has the scale, translation, rotation invariance, but also the use of more stable and faster image retrieval.