A New Image Retrieval Method Based on K-Nearest Neighbor Multistage and Multiple Features

摘要

This paper proposes a new image retrieval method based on the fuse of color, texture and edge features. The color feature is extracted from images by HSV histogram weighted sub-block. The texture feature is extracted by the Local Binary Pattern (LBP) algorithm. The edge as the image shape feature is extracted by the Canny operator. A similarity computing algorithm based on K-Nearest Neighbor is proposed to carry out the multistage similarity compute. The extracted features such as color, texture and shape are used for the multistage similarity compute. The retrieval results are ordered by the similarity values. The experimental results show that the new image retrieval algorithm has higher retrieval precision than single image feature retrieval.