A Color Image Clustering Method Based on Non-negative Tensor Ring

Abstract

The invention provides a color picture clustering method based on a non-negative tensor ring, comprising the following steps: constructing an error function of the non-negative tensor ring decomposition, setting an iteration termination condition; setting the rank of the non-negative tensor ring decomposition; solving the error function, when When the termination iteration condition is reached, all updated tensor cores are output; specific tensor cores are selected for matrixing, and non-negative low-dimensional data features of color pictures are extracted for clustering. The non-negative tensor ring-based color picture clustering method of the present invention introduces tensor ring decomposition and non-negative constraints in dimension reduction and feature extraction, eliminates the restriction on rank, and effectively extracts non-negative low-dimensional data features of color pictures for clustering class, which improves the clustering effect of color image datasets.

Description

technical field [0001] The invention relates to the field of image clustering, in particular to a color image clustering method based on a non-negative tensor ring. Background technique [0002] With the development of the times, we have entered the era of big data. With the advancement of technology, data collection has become easier and easier, resulting in increasing data scale and complexity, such as various types of trade transaction data, Web documents, gene expression data, document word frequency data, User rating data, WEB usage data and multimedia data, etc., their dimensions (attributes) can usually reach hundreds of thousands of dimensions, or even higher. Usually a two-dimensional data is called a two-dimensional tensor (matrix), and high-dimensional data is collectively called a high-dimensional tensor. In the natural data, such as images, videos, gene expression, user ratings, etc., are all non-negative high-dimensional data, so there is an urgent need for e...

AI Technical Summary

Problems solved by technology

[0006] The present invention provides a color picture clustering method based on non-negative tensor rings, which solves the problem of not being able to perform effective feature extraction on color pictures in the prior art. The present invention can ensure that the original information is not lost, Maximally break through the bottleneck to improve the performance of color image clustering

Images