Cross-quality face recognition method based on convolutional neural network features

Abstract

The invention provides a cross-quality face recognition method based on convolutional neural network features. The method comprises the steps that firstly, acquiring image blocks of feature points ofa high-quality training sample image, a low-quality test sample image and a high-low quality training dictionary sample image; secondly, designing a deep convolutional neural network, and for each feature point image block, obtaining a feature vector through learning of a neural network; performing linear representation on the feature vector of test image blocks and a feature vector of training image blocks; performing similarity measurement on the feature representation of low-quality test image blocks and the feature representation of the high-resolution training image blocks, and outputting the category of each test image block; and finally, dividing one face image into an image block set of S face key points, voting the image block classification result of each key point position, distributing the image to the class with the maximum vote number, and outputting the class of the final low-quality test image.

Description

technical field [0001] The invention relates to an image recognition method, in particular to a cross-quality face recognition method based on convolutional neural network features, belonging to the technical fields of pattern recognition and biological feature recognition. Background technique [0002] Face recognition technology is a popular research topic based on computer, image processing and pattern recognition. For a long time, with the wide application of face recognition in various social fields, such as criminal case identification, public security system, monitoring, etc., face recognition technology has received more and more attention. [0003] In the process of face recognition, there is a problem of low recognition accuracy due to inconsistent face image quality, and sometimes it is difficult to complete the recognition work. Existing face detection methods: [0004] [1] X.Cao, Y.Wei, F.Wen, J.Sun, "Face alignment by explicit shape regression" Int.J.Computer...

AI Technical Summary

Problems solved by technology

[0007] The disadvantage of the existing method is that it cannot process the recognition between images of different quality in a timely manner, and the recognition efficiency will be greatly reduced due to factors such as illumination and occlusion.

Images