A Single-Sample Face Recognition Method Based on Adaptive Virtual Sample Generation Criteria

Abstract

The invention discloses a single-sample face recognition method based on an adaptive virtual sample generation criterion, which mainly solves the problem of low face recognition rate in the prior art. The implementation steps are: 1. Select the face image and divide the training and test sample sets; 2. Perform singular value decomposition on the training sample, and reconstruct a new training sample image according to the decomposed base image; 3. Combine the training sample image and The new reconstructed image constructs a virtual training sample image, and divides the training sample image and the virtual training sample image into blocks to form a block training sample set; 4. Use these block training samples to train the optimal projection space; 5. Use the same method to The test sample is divided into blocks, and projected to the optimal space to obtain the block sample characteristics; 6. Classify the block test samples according to the block sample characteristics, and use the maximum voting criterion to obtain the final recognition result. The invention reduces the lack of identification information in face recognition, improves the face recognition rate, and can be used for identification of ID cards, driver's licenses and passports.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a face recognition method, which can be used for identification of ID cards, driver's licenses and passports. technical background [0002] Face recognition is a hot topic in the fields of pattern recognition and computer vision. In recent years, many methods have been proposed and widely used in public security and video surveillance. But it is also a difficult and complicated problem. For example, considering the problem of sample storage and the difficulty of sample acquisition, it is often faced with the situation that there is only one training sample for each type. In this case, some commonly used face recognition methods cannot For direct application, it is necessary to design a recognition algorithm that can effectively extract the essential discriminative features of different individuals from a single training sample. Therefore, designing an effective single tr...

AI Technical Summary

Problems solved by technology

However, both methods have a common disadvantage that the number of base images in the reconstruction process is fixed, resulting in the loss of some identification information, which affects the recognition rate of the face recognition system.

Images