Non-negative image data dimension reduction method based on Hessian regular constraint and A optimization

Abstract

The invention discloses a non-negative image data dimension reduction method based on Hessian regular constraint and A optimization. The method includes the steps that step1, a sample feature matrix is constructed; step2, a Hessian regular matrix is calculated; step3, a basis matrix and a coefficient matrix are iteratively output so as be analyzed in a clustered mode. An A optimization regular item and a Hessian regular item are added into a target function so that a data expression obtained through decomposition can keep inherent features included in the manifold-shaped structure of original data while guaranteeing a small prediction error; through dimension reduction, redundancy information in the high-dimension data is removed, the low-dimension expression capable of accurately expressing the semantic structure of the data is extracted, and therefore clustering analysis performed on the high-dimension data becomes simpler and more effective.

Description

technical field [0001] The invention belongs to the technical field of image data processing, and in particular relates to a non-negative image data dimensionality reduction method based on Hessian regular constraints and A optimization. Background technique [0002] In image processing fields such as image clustering, image recognition, and image-based scene pattern recognition, the scale of data has exploded recently, and these image data to be processed often have very high-dimensional features. These massive high-dimensional feature data have brought many challenges in image storage and processing. Fortunately, the researchers discovered that the intrinsic dimensions of these image data are much lower than their original dimensions. This practice of using low-dimensional data to replace the original high-dimensional representation of data is called dimensionality reduction. For many data analysis methods, the use of dimensionality reduction can reduce the constraints o...

AI Technical Summary

Problems solved by technology

Therefore, the basis obtained by these non-negative matrix factorization methods may be far away from the original data, and it is obviously not optimal to use such a basis for data representation.

Images