Font Size: a A A

Low Resolution Face Recognition System Design Based On FPGA

Posted on:2018-04-28Degree:MasterType:Thesis
Country:ChinaCandidate:B LiFull Text:PDF
GTID:2348330533465864Subject:Microelectronics and Solid State Electricity
Abstract/Summary:PDF Full Text Request
Face recognition has been a hot research direction in artificial intelligence and computer vision fields for a long time, and put into use in many areas. It's very simple for a normal eyesight person to recognize a familiar face, but hard for the blind. Face recognition system is the best way to help the blind person in promoting their recognition rate and living quality because facial features are significant difference to everybody.According to be a recognition assistant system ,the circumstance of blind person living and the visual prosthesis program, a low resolution face recognition system used in simple environment has been designed in this paper, the system is verified on the FPGA board. After the comprehensive analysis of the recognition effect of all low resolution face recognition algorithms in the simple environment, the PCA and LDA algorithms have been selected in the paper because of their about 95% recognition rate in the simple face library.Recognition module circuit has been also designed in this paper based on the principle and learning result of the algorithm. The circuit could be divided into three modules including the feature extraction module, the Euclidean distance computation module and the minimum Euclidean distance computation module. After all circuit design, the recognition circuit has been verified and tested on the FPGA board, the test result shows that the circuit can recognize the test low resolution faces correctly, and the speed of recognition circuit is much faster than software at about 20us.In order to make the recognition circuit more easily applying to the recognition assistant system for the blind person, a system has been build used SOPC technology in this paper and the recognition circuit module is connected to the system by Avalon bus. The system read the high resolution face image and then displayed it by VGA module according to the result of recognition circuit. The experiment show that the system needs about 0.16s to recognize and display the face image except the system start-up first time?Based on the comprehensive analysis of the recognition effect of all low resolution face recognition algorithms in the simple environment, the obvious advantage of PCA+LDA algorithm has been found through the theory and experiment in this paper, and the work of this paper has some reference value in the similar research in future.
Keywords/Search Tags:low resolution, face recognition, PCA, LDA, FPGA
PDF Full Text Request
Related items