Fuzzy face image verification method based on phase-encoded features and multi-metric learning

Abstract

The invention discloses a fuzzy face image verification method based on phase encoding features and multi-metric learning, including: (1) training stage: block the sample image, and extract multi-scale primary features for each image block; use the above feature to learn the fisher kernel dictionary to generate block fisher kernel encoding features; perform multi-metric matrix learning on the above-mentioned encoded features to generate multiple metric matrices, and obtain the metric distance of the training samples after multi-metric matrix projection, and calculate the positive samples, Negative samples measure the average distance and variance of the set respectively, and determine the final classification threshold through the probability calculation formula of Gaussian distribution; (2) Verification stage: For the input face image, divide the image into blocks and extract multi-scale primary features, Then generate block fisher kernel encoding features, and then obtain the final measurement distance through the multi-metric matrix, and compare this distance with the threshold to obtain the face verification result. The invention has the advantages of high recognition rate and strong versatility.

Description

technical field [0001] The invention relates to the fields of computer vision and pattern recognition, in particular to a fuzzy face image verification method based on phase encoding features and multi-metric learning. Background technique [0002] Face recognition and verification technology has been a research hotspot in the field of computer vision and pattern recognition in the past few decades, and it is also widely used in intelligent monitoring, identity verification and other occasions. After decades of development, face recognition and verification technology has a very high accuracy rate in a controlled environment, but in real applications there are many factors that will affect the accuracy of face recognition and verification, image blur and resolution Low is one of the very important influencing factors. [0003] The main reasons for image blur are as follows: 1. When extracting faces from urban surveillance videos for identification and verification, the obta...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problem that the existing face recognition and verification methods cannot robustly deal with common blurred and low-resolution face images in the real environment, the present invention proposes a fuzzy face image verification method based on phase encoding features and multi-metric learning , the method can extract compact and descriptive anti-blur features from blurred face images, and combined with the proposed block measurement method, it improves the classification accuracy of the verification algorithm and compresses the data dimension. On the real data, it still has a good recognition accuracy for blurred images

Images