The Study of Adaptive Filters for Image Noise Cancellation

• Main Authors: Chia-Feng Yu, 喻佳風
• Other Authors: Chian C. Ho
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/88969772865440191756

碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 95 === In this thesis, four modified TDBLMS algorithms for image noise cancellation are proposed. We classify the algorithms into two categories: the first category includes that developed based on the modified fast convergence BLMS algorithms; the second category includes the adaptive filters with the modified weight-training mechanism. The modified fast convergence mechanism speeds up the convergence and improves the stability in the application of image noise cancellation. The modified weight-training mechanism shortens the time required to obtain the suitable weight coefficients for further processing. In addition, the computational complexity is reduced. During simulation, we take the images with four different block sizes to verify the proposed algorithms and compare with the existing three adaptive algorithms. The experimental results indicate that the proposed algorithms reduce noise effectively and generate the better visual effect. Moreover, they improve the PSNR with 6%~36% and reduce the computational complexity with 33% comparing with the conventional ones.