The VLSI Design For Visual Prosthesis Wavelet Edge Detection Algorithm

Vision is the most important of human sense organs.The study found that most of the blind are just part of the lesions occur in the visual pathway,this study laid a physiological basis to the research of visual prosthesis.Visual prosthesis technology is a kind of function point stimulation technology,By implanted microelectric array to the eyes of the blind person to stimulate its unscathed is part of the visual pathways,generate phosphene.But the current visual prosthesis technology is subject to the instability of low resolution,phosphene and phosphene delay problems' limit,This needs through image processing under a small number of pixels to the main information real-time object recognition for blind people.Edge is the most important part of the image information,Through to the edge image recognition can guarantee blind person'basic visual perception.And the wavelet analysis is emerging in recent years popular signal processing method,Its good time-frequency characteristics of local analysis and multi-scale multi-resolution ideas with high application value in the image processing.In this paper,based on the theory of wavelet transform,designed wavelet edge detection algorithm.Design based on FPGA is to focus on lifting wavelet transform VLSI circuit structure,proposes a parallel computing circuit of 9 input,guarantee a small amount of pixel edge continuity,And join threshold interception in the high frequency operation per level transform extract more elaborate edge curve.The thesis firstly design algorithm on MATLAB software and compare each kind of edge detection algorithm,analyzes the effect of algorithm is introduced.Then design of hardware circuit and the simulation algorithm modules for the RTL design in detail,at last,by CMOS camera,SDRAM memory,VGA display set up video processing system verification was carried out on the calculation in system level.Improve design in this paper wavelet arithmetic belongs to the operation of the integer domain,same address calculation,compared with the traditional wavelet faster,easy to reconstruct the characteristics of the suitable for hardware implementation,and in the process of video stream processing about time-sharing multiplexing,detect more accurate edge information and can meet the demand of visual prosthesis is the size of the resolution of video real-time transmission,enrich the research contents of visual prosthesis image processing.