A Face Recognition Method, Device and Equipment Based on Neighbor Preserving Low Rank Representation

Abstract

The embodiment of the present invention discloses a face recognition method, device, equipment and computer storage medium based on neighbor-preserving low-rank representation. Among them, the method includes integrating similarity adaptive preservation and low-rank representation into a unified learning framework, obtaining the representation coefficient matrix and extracting features, based on kernel paradigm and L 2，1 ‑Sparse projection matrix under paradigm constraints; use the sparse projection matrix to extract the salient features of the training sample set and test sample set similarity adaptive preservation, and generate the facial feature training sample set and facial feature test sample set embedded with salient features ; Use the face feature training sample set to integrate and optimize the reconstruction error minimization item representing the coefficient matrix and salient features; input the face feature test sample set into the nearest neighbor classifier constructed by the face feature training sample set, and according to the similarity Identify and get the identification result. The technical solution provided by this application improves the feature extraction and recognition capabilities of face images.

Description

technical field [0001] Embodiments of the present invention relate to the technical fields of computer vision and image recognition, and in particular to a face recognition method, device, device and computer storage medium based on neighbor-preserving low-rank representation. Background technique [0002] With the increase of data volume and content complexity, how to effectively and robustly represent data has become more and more important in the field of data mining and analysis. Currently, data representation methods include dimensionality reduction, sparse representation, and low-rank recovery. [0003] In the low-rank representation model, robust principal component analysis (RPCA) and low-rank representation (LRR) are generally used. Robust PCA and low-rank representations aim to decompose a given data matrix into a low-rank component and a sparse error part, the low-rank part being equivalent to a compact representation of the original data. Due to low-rank encodi...

AI Technical Summary

Problems solved by technology

However, most of the current low-rank coding methods cannot effectively preserve domain information or local geometric structure of data.

Images