A Multi-view Data Subspace Clustering Method

Abstract

A multi-view data subspace clustering method is disclosed, which can fully reveal the structural information of the multi-view data and improve the clustering performance of images. This multi-view data subspace clustering method comprises the following steps: (1) adding low-rank representation constraints in the multi-view data subspace clustering method; (2) adding a multi-view data subspace clustering method Add sparse representation constraints in ; (3) Construct a specific form of clustering similarity matrix through a cascade before and after.

Description

technical field [0001] The invention relates to the technical field of image classification, in particular to a multi-view data subspace clustering method. Background technique [0002] With the rapid development of science and technology, the ways of obtaining data are more diversified. Massive text, image, audio and video data play an important role in all aspects of people's lives. The analysis and processing of large-scale data is increasingly important in scientific research. domain occupies a pivotal position. The high complexity of data content leads to increased data dimensions, and a lot of data can be observed from different sources or described by various types of features. For example, surveillance videos can obtain information on the same location from different angles, and images can use different features. Description (such as local binary features, texture features, spatial envelope features, etc.). In fact, these data can be regarded as signals observed by...

AI Technical Summary

Problems solved by technology

[0005] However, the current multi-view data clustering methods usually only consider the low-rank or sparse constraints on the similarity matrix, and do not make full use of the gains of the two constraints on the self-representation model.

Images