Image Clustering Method Based on Sparse Orthogonal Dual-Graph Non-negative Matrix Factorization

Abstract

The invention proposes an image clustering method based on sparse orthogonal dual-image non-negative matrix decomposition, which is used to solve the technical problems of low accuracy and slow speed of image clustering existing in existing methods. The implementation steps are: input image data; calculate data space similarity matrix and feature space similarity matrix; calculate data space similarity diagonal matrix and feature space similarity diagonal matrix; obtain label constraint matrix; define and initialize three sparse normal Intersecting two-graph non-negative matrix factorization factor matrix; set the number of iterations; obtain the update formula and label constraint matrix update formula of three sparse orthogonal two-graph non-negative matrix factorization matrix; define the coefficient diagonal matrix update formula; A sparse orthogonal double-image non-negative matrix factorization factor matrix, label constraint matrix and coefficient diagonal matrix are updated; definition and calculation of low-dimensional data representation matrix; image clustering and output. The invention can be used in practical applications such as text, image clustering and face recognition.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to an image clustering method, in particular to an image clustering method based on sparse orthogonal dual-image non-negative matrix decomposition, which can be used for text clustering, image clustering and face recognition and other practical applications. Background technique [0002] With the development of shooting tools such as cameras, video cameras and smart phones, images have become indispensable data information in people's lives, and image clustering methods, as an effective data organization method, have therefore become one of the hottest research directions. Image clustering is to divide the image data into several categories according to the number of selected categories, so that the intra-class similarity in the same class is as large as possible, while the inter-class similarity between different classes is as small as possible. Commonly used image clusteri...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problems existing in the above-mentioned existing methods, and propose an image clustering method based on sparse orthogonal dual-image non-negative matrix decomposition, which is used to solve the problems of low image clustering accuracy and The technical problem of slow clustering

Images