Study On Algorithm Of Wavelet Domain Image Denoising By Thresholding And Wiener Filiter Base On FPGA

As an important component of signal processing, image processing has played a more and more important role in information society. However, image is often interfered by noise in collection and transmission. In the further analysis and processing, we must reduce noise level in image preprocessing. As the wavelet transform with low-entropy, multi-resolution, de-correlation and other features, it has been widely used in image denoising, and then it shows a great prospect of application in recent years.First, this dissertation presents the basic theory of wavelet transform, the basic idea of multi-resolution analysis and the basic methods of wiener filter. We mostly focuses on wavelet domain image denoising by thresholding, and combines it with wiener filter which is the most optimal filter on means of square error sense, then educes a method of wavelet domain image denoising by thresholding and wiener filter. The simulation tests indicate the algorithm is effective and available under Gaussian noise.Based on the above work, this dissertation gives the design of hardware in the platform using Xilinx's Virtex4 XC4VLX80 FPGA. According to the specialty of image denoiseing, we devise the hardware construction, which has 5 pipeline stages including input, wavelet transform, wiener filter, inverse wavelet transform, output; and then exchange data through out-chip memory. Each pipeline stage has to work in 1 frame, independent conducts its operations, and it can avoid the waiting time. This paper describes the design of every module in details, devises the hardware construction for wavelet transform in the practical engineering. The hardware is coded by Verilog HDL, and it has passed the system testing.Overall, after a lot of theory study and sufficient understanding, this project focuses on concrete application research and gets several achievements.