Color Face Recognition Method in Hypercomplex Encrypted Domain

Abstract

The present invention proposes a color face recognition method in a super-complex encrypted domain, comprising the following steps: using a two-dimensional Logistic adjusted sinusoidal mapping to generate a double random phase mask; representing the color face image to be recognized as a pure quaternary matrix, and Encrypt according to the double random phase mask and quaternion Fourier transform; extract the real and imaginary components of the ciphertext data, and combine the sparse matrix to obtain the sparse ciphertext components, and obtain the sparse ciphertext according to the ciphertext components and decrypt it; construct the quaternion minimum average correlation energy average filter according to the invisible color face image obtained by decryption; perform quaternion inverse Fourier transform on the product of the color face image to be identified and the filter, And calculate the peak side lobe ratio PSR value, and use the category of the filter corresponding to the maximum PSR value as the category of the color face to be recognized. The invention can effectively improve the recognition accuracy, and can hide the content of the face image, and has higher security.

Description

technical field [0001] The invention relates to the technical field of biological feature recognition, in particular to a color face recognition method in a hypercomplex encryption field. Background technique [0002] As an important aspect of biometric identification, face recognition has been widely studied and applied in the fields of computer vision monitoring, human-computer interaction and identity authentication. In general, face recognition methods include two stages of feature extraction and recognition. [0003] In an unconstrained natural environment, face images inevitably have problems such as scale scaling, angle rotation, perspective transformation, occlusion, and illumination. In order to improve the face recognition rate in the harsh environment mentioned above, some face images based on correlation The recognition method has been continuously developed, and the design of this filter is mainly based on the face images of the training set in the Fourier tran...

AI Technical Summary

Problems solved by technology

Therefore, information security has become a crucial issue, and the current relevant face recognition method is to directly use the original color face image and the quaternion minimum average correlation energy mean filter for recognition. The recognition accuracy and security of this method The performance is not high enough to meet the needs of users

Images