FPGA-based Fast Wavelet Denoising System For Scanning White Light Interference Data Of Fracture Surface Topography

The detection of the microtopograthy of the fracture surface and its quantitative analysis are of great value for the study of the generation and expansion of fractures,which has received more and more attention.When the microtopograthy of the fracture surface is detected by the scanning white-light interferometry,the quality of the detection data is inferior due to the complicated physical condition of the surface.As result,it will bring great challenges to subsequent processing and reconstruction and consume a lot of time and space resources.In order to improve the work efficiency of system and consider the real-time detection in the future,it is very important to realize the rapid data processing.Combining with the development of FPGA technology in recent years and the unique advantages of this technology in signal processing,the research on fast denoising of scanning white-light interference data of fracture surface topography is proposed and performed.In this work,a fast data processing system,which is mainly composed of FPGA hardware and LabVIEW data interface software,is designed.The data to be processed are read from the PC and then sent to the hardware module of FPGA.After the processing by hardware,the processed data are transmitted back to LabVIEW interface and save to PC.Furthermore,the FPGA consists of DDR2 memory module,wavelet denoising module and USB2.0 interface and control.The LabVIEW interface software is divided into three parts: data transmission,reception and access.Meanwhile,the functions of path traversal,image processing effect contrast display and time consuming monitor are also designed based on the LabVIEW soft.The Altera's Cyclone IV series chip EP4CE22F17C8 N is selected as the main controller for the FPGA system,and each module is designed based on the Verilog hardware description language.In this system,the advantages of parallel processing of FPGA,the characteristics of high-speed reading and writing of DDR2 and the application of discrete wavelet algorithm have significantly improved the efficiency of data processing.An improved 5/3 lifting wavelet transform algorithm is used to filter the white-light interference image,so that the image and edge data are denoised uniformly and the processing effect is improved.The experimental results show that the FPGA-based image processing system can realize fast and batch processing of white-light interference images on the fracture surface.The analysis on the experimental results shows that the average PSNR before and after denoising is 44.291 dB and 52.218 dB,respectively,and the former is increased by 7.927 dB compared with the later,while the MSE is reduced by 2.031,achieved better denoising effect.According to LabVIEW 's measurement,it takes about 8ms for the system to process an image,and the FPGA hardware takes only about 107.76 ?s,which is far less than the processing time of 2.77 s by Matlab software under the same operating environment.Thus,the timeliness of the system is proved.