An Unconstrained Face Image Dimensionality Reduction Method Based on Discriminative Sparse Preserving Embedding

Abstract

The present invention provides an unconstrained face image dimensionality reduction method based on discriminative sparseness preserving embedding. By 1) calculating the sample reconstruction relationship matrix W, when calculating the sample sparse reconstruction relationship, class labels are introduced to construct intra-class reconstructions respectively The relationship matrix and the inter-class reconstruction relationship matrix, and increase the intra-class and inter-class compactness constraints in the sparse reconstruction stage, effectively increase the reconstruction relationship between the samples to be tested and similar samples, and weaken the overlap between samples to be tested and heterogeneous samples. 2) Calculate the low-dimensional projection matrix P, when calculating the low-dimensional projection matrix, increase the global constraint factor, not only consider the local sparse relationship of samples, but also consider the global distribution characteristics, and further weaken the impact of heterogeneous pseudo-nearest neighbor samples on low-dimensional projection 3) Realize the low-dimensional linear mapping of high-dimensional sample data; this method greatly improves the accuracy of face recognition in an unconstrained environment.

Description

technical field [0001] The invention relates to an unconstrained face image dimensionality reduction method based on discriminant sparseness preserving embedding, which uses face recognition in an unconstrained environment as the application background to conduct research on low-dimensional mapping of high-dimensional face data, mainly including sample sparseness Optimization of structural relations and improvement of low-dimensional projection objective functions. Background technique [0002] With the rapid development of the Internet and sensor technology, the facial image data processed by computers is becoming more and more massive and complex. Therefore, it is particularly important to effectively reduce the dimensionality of massive complex face data and dig out useful essential information hidden under the high-dimensional appearance. On the one hand, it can reduce the data dimension, save storage space, and improve the operating efficiency of the system; on the oth...

AI Technical Summary

Problems solved by technology

[0009] 1. LPP, NPE and their improved algorithms are linear implementations of traditional manifold learning algorithms, which effectively maintain the local neighbor relationship between samples, but in high-dimensional space, describing the neighbor relationship with the Euclidean distance between samples does not have Distinguishability, therefore, expressing the neighbor relationship with the distance measure cannot accurately mine the useful low-dimensional essential information hidden in the high-dimensional redundant data

[0010] 2. In the locally preserving dimensionality reduction algorithms represented by LPP and NPE, the construction of the nearest neighbor graph plays a crucial core role, but the selection of appropriate graph parameters is still a technical difficulty for this type of algorithm. The tiny neighbor graph Any parameter change will lead to a sharp decline in the performance of the algorithm.

However, when dealing with face data acquired in an unconstrained environment, the samples are complex and changeable. Traditional SPP is an unsupervised dimensionality reduction method, and the optimized sparse reconstruction weights cannot accurately reflect the identification information. Moreover, in low-dimensional When projecting, the global distribution characteristics of the samples are not considered, which affects the accuracy of unconstrained face recognition

Images