A face recognition method and system based on integral method

Abstract

The invention discloses a face recognition method and system based on the integral method. The system includes a first normalization module, a first convolution module, a first low-pass filter connected in sequence, and a second normalization module connected in sequence. Module, the second convolution module, the second low-pass filter, the first low-pass filter and the second low-pass filter are respectively connected to the input of the first multiplier, and the output of the first multiplier is connected to the third low-pass filter, the third low-pass filter is connected to the input terminal of the integrator, and the output terminal of the integrator is connected to the comparator. The method is as follows: respectively perform normalization processing, convolution scanning and low-pass filtering processing on the face to be recognized and the reference face, multiply the processing results and perform low-pass filtering and integration, and compare the integration results with the preset threshold value Compare and get the recognition result. This design is based on simple computing methods and devices, without the need for huge calculations and complex model construction, it can complete high-precision recognition of human faces, with high recognition efficiency and low system cost.

Description

technical field [0001] The invention relates to the field of face recognition, in particular to a face recognition method and system based on an integral method. Background technique [0002] Face recognition technology, as a high-precision identification technology, is widely used in banking, security, security inspection, security, and escape pursuit. The traditional face recognition method is to use a large number of face pictures to generate a neural network after training, and then use the neural network to generate a feature code, and perform a 1:1 comparison on the feature code. When the comparison threshold exceeds the set threshold, it is judged as the same face. This method requires a large number of pictures and computing power to generate a relatively high-precision recognition model when training a neuron network, and the model parameters are as many as one million, and the amount of calculation is extremely large, requiring the cooperation of high-performance ...

AI Technical Summary

Problems solved by technology

This method requires a large number of pictures and computing power to generate a relatively high-precision recognition model when training a neuron network, and the model parameters are as many as one million, and the amount of calculation is extremely large, requiring the cooperation of high-performance calculators. Trained by highly skilled engineers, the recognition effect is unsatisfactory for models with poor training results

Images