Method and device for face verification

Abstract

The invention discloses a method and a device for face verification, which belongs to the field of face recognition. The method comprises the steps: acquiring matching points between a first face image and a second face image; partitioning the first face image and the second face image when the quantity of the matching points is greater than a preset quantity threshold value; counting the quantity of the matching points in each block of the first face image and the second face image, and obtaining a first matching vector and a second matching vector; calculating similarity between the first matching vector and the second matching vector; verifying whether the first face image is matched with the second face image according to the similarity. The device comprises an acquisition module, a partitioning module, a statistics module, a calculation module and a verification module. By adopting the method and device, no restriction for the to-be-verified face image is needed, a better compatibility for the image outside a sample library can be realized, and the application range is wider.

Description

technical field [0001] The invention relates to the field of face recognition, in particular to a face verification method and device. Background technique [0002] Face verification is a branch of the field of face recognition. Face recognition includes two categories: ① face identification: identify the identity of a person based on the face image, and solve the problem of who is it, which is a one-to-many matching process; ② face verification: judge the face in the face image It is a one-to-one matching process to solve the problem of whether it is the designated person or not. Most of the traditional face verification algorithms still use Principal Component Analysis (PCA) for dimensionality reduction to obtain face features, and then use Linear Discriminant Analysis (LDA) to obtain inter-class differences. To determine whether the face matches, it belongs to the machine learning algorithm of the supervised training process. [0003] At present, there is a face verifi...

AI Technical Summary

Problems solved by technology

However, the face verification algorithm needs training, and the verification must be supported by a trained database to complete the verification. It is impossible to compare external images that have not been trained, which has great limitations.

Images