An Improved Method for Image De-Noising Based on Lifting Scheme
Image de-noising is a classical topic in the field of signal analysis and image processing, and is also a fundamental basis for image analysis and interpretation, but until now, the problem has not been solved well and is a challenge for scientists and engineers. As a powerful tool, wavelet transform plays an important role in signal analysis, image processing and computer vision. De-nosing of natural images corrupted by Gaussian white noise using wavelet transform is an effective way because of its ability to capture the energy of a signal in few energy transform values. Additive random noise can easily be removed using simple thresholding methods such as hard or soft thresholding. However, these classical de-nosing methods in wavelet domain have their limitations. In this paper, the author briefly present the lifting scheme and its main steps, and then introduces the theory of traditional wavelet thresholding methods in image denoising and their defaults. At last, we propose a new improved method for image de-noising based on lifting scheme, the new method estimates the standard noise variance, which calculates the energy of each coefficient in a neighbourhood and brings it into the shrinkage scheme. Experiments demonstrate the feasibility of the new method.
Image de-noising wavelet transform lifting scheme thresholding wavelet coefficients
Haiyang Wu Hui Wang Wen An
Department of Remote Sensing, Institute of Surveying and Mapping, Information Engineering University Department of Remote Sensing, Institute of Surveying and Mapping, Information Engineering University
国际会议
2010 International Conference on Image Analysis and Signal Processing(2010 图像分析与信号处理国际会议 IASP 10)
厦门
英文
56-60
2010-04-12（万方平台首次上网日期，不代表论文的发表时间）