Face identification method based on combined partitioning and system thereof

Abstract

The invention discloses a face identification method based on combined partitioning and a system thereof. The method comprises the following steps of carrying out partitioning processing on an acquired face image and acquiring a partitioned face image partitioning composition class set; carrying out k-mean algorithm cluster processing on the face image partitioning composition class set and converting a cluster result into a cluster matrix; marking the face image partitioning composition class set, calculating distances among face image blocks, and acquiring the two face image blocks with a shortest distance in each set; according to the two face image blocks with the shortest distance in each set, expanding the cluster matrix, according to an expansion result, carrying out class sparse expression calculation, and acquiring a calculated combined partitioning dictionary matrix; carrying out reconfiguration error minimum classification processing on the calculated combined partitioning dictionary matrix, according to a reconfiguration result, carrying out face identification and acquiring a face identification result. In the invention, face identification can be rapidly and accurately realized.

Description

technical field [0001] The invention relates to the technical field of biometric feature recognition, in particular to a face recognition method and system based on joint segmentation. Background technique [0002] Since the 1990s, face recognition has become a popular research field in computer vision, and the research on face recognition has achieved many important results. In recent years, the focus of face recognition research has shifted to face recognition in an unrestricted environment, including human pose, expression, face size, face occlusion, distance from the face to the camera, and the surrounding environment. Environmental light changes, etc. In recent years, many methods have been proposed to solve the above problems. [0003] Algorithms based on sparse representation classification (SRC, sparse representation classification) have been widely proposed recently. In the method of sparsity representation, a dictionary is constructed with a standard image set, ...

AI Technical Summary

Problems solved by technology

In addition, images of different classes acquired in the same environment may be more correlated than images of the same class acquired in different situations

Therefore, the unsupervised 1-norm minimization of sample coefficients in SRC may cause the representation coefficients of the test image to come from many different categories, which will lead to classification errors

[0006] Traditional face recognition algorithms such as SRC, etc., if you want to achieve a high face recognition accuracy rate, there are many restrictions on the conditions, and the amount of calculation is large, the recognition time is long, the efficiency is low, and the recognition accuracy rate needs to be improved.

Images