Face recognition method based on multi-patch and multi-channel joint feature selection learning based on CNN

Abstract

The invention discloses a face recognition method based on CNN-based multi-patch and multi-channel joint feature selection learning, belonging to the technical field of face recognition. This method first divides the original face image into multiple sub-images, and then separates each sub-image into multiple channel images; then constructs a CNN network model for each channel image, and inputs the channel image for recognition; The multiple channel neural networks of the sub-images are connected to obtain multiple sub-image neural networks corresponding to multiple sub-images, and then the multiple sub-image neural networks are connected as the final model recognition result. The present invention improves and innovates the existing convolutional neural network model, so as to achieve the effect of optimizing and upgrading the model, making the face recognition ability of the convolutional neural network model more accurate, and its application in daily life, industrial development, and scientific research Wide range of applications in other fields provide stronger technical support.

Description

technical field [0001] The invention belongs to the technical field of face recognition, and in particular relates to a face recognition method based on a convolutional neural network. Background technique [0002] In recent years, biometric-based identification technology has been widely used in many scenarios of daily life. Among many biometric technologies, face recognition technology has the advantages of non-invasiveness, non-contact, and ease of operation, and the collection of face image data is also easier. This also makes the application scenarios of face recognition technology in information security, identity verification, site monitoring, human-computer interaction and other fields more extensive. Therefore, in-depth research on face recognition has important theoretical and practical significance for attendance, security, entertainment and other aspects. [0003] At present, common face recognition methods mainly include: face recognition methods based on geom...

AI Technical Summary

Problems solved by technology

Although the current face recognition method based on deep learning is better than many traditional algorithms in terms of accuracy, it also has some shortcomings. For example, this method often ignores some local key features of the face, and only Feature learning will be performed on the overall face of the original image, which reduces the accuracy of face recognition to some extent

Images