Image Recognition Based on Directed Complex Network Model

Abstract

With the rapid development of computer and information technology, the applications of the image recognition become more and more universal. As the basis of the image processing, the image representation plays an increasingly important role in computer vision and image recognition. In recent years, complex network theory has aroused the interest of many researchers. Most of the complex network features of images are statistical features, which not only have good stability, but also have strong anti-noise ability. A directed complex network representation model is proposed in this paper, which uses K-Nearest Neighbor (KNN) method to give an evolution method of directed complex networks. Finally, the feature description of the image is completed and the recognition of the image is realized by extracting the directed complex network features at different evolutionary times.