Non-negative feature representation and recognition method, device and system for face data in self-constructed cosine kernel space and storage medium

Abstract

The invention provides a face data non-negative feature representation and recognition method, device and system in a self-constructed cosine kernel space and a storage medium. The face data non-negative feature representation and recognition method in the self-constructed cosine kernel space comprises a training step and a recognition step. The beneficial effects of the invention are that the method carries out the experiment comparison with a related algorithm in a public face database, and a result shows that the method has certain superiority; through experimental comparison with a relatedalgorithm in a face database added with noise, the result shows that the method has good robustness.

Description

technical field [0001] The invention relates to the technical field of face recognition, in particular to a non-negative feature representation and recognition method, device, system and storage medium of face data in a self-constructed cosine kernel space. Background technique [0002] With the advent of the information age, biometrics, which uses the inherent physiological and behavioral characteristics of the human body for personal identification, has become one of the most active research fields. Among the many branches of biometric technology, the most easily accepted technology is face recognition technology, because compared with other biometric technologies, face recognition is non-invasive, non-mandatory, and non-contact. and concurrency. [0003] Face recognition technology consists of two stages. The first stage is feature extraction, which is to extract the face feature information in the face image. This stage directly determines the quality of face recognitio...

AI Technical Summary

Problems solved by technology

However, most of these KNMF algorithms have the following problems: (1) the analytical expression of the nonlinear mapping implicit in the kernel function cannot be obtained; (2) the mapped data cannot be guaranteed to be non-negativity in the kernel space, so the current The KNMF algorithm can only be regarded as a semi-nonnegative matrix decomposition; (3) requires imprecise pre-image learning; (4) is not robust to noise

[0035] 1. The non-negative matrix factorization algorithm is a linear algorithm, but many problems in real life are nonlinear, so it is difficult to achieve satisfactory results

[0036] 2. The current kernel non-negative matrix factorization algorithm generally uses polynomial kernel function or Gaussian kernel function, but it is difficult to obtain the analytical expression of the nonlinear mapping implied by it

In addition, the current KNMF algorithm only uses the first three terms of the Taylor expansion to solve the original image W, so the error is large, and its original image learning is inaccurate

An imprecise preimage can affect its performance

[0037] 3. The kernel method based on polynomial kernel function or Gaussian kernel function cannot guarantee the non-negativity of the mapped data, in fact, it is a semi-non-negative matrix decomposition

[0038] 4. Most of the current kernel non-negative matrix factorization algorithms are based on polynomial kernel functions or Gaussian kernel functions. These two kernel functions are more sensitive to noise, which makes the algorithm less resistant to noise.

Images