Abstract:
In recent years, the wavelet-based image denoising techniques mainly focus on the case that uses a single wavelet base. In fact, different wavelet bases can often effectively represents different areas or features of an image. In this paper a new image denoising method using multiple wavelet bases is proposed. First, the noisy image is denoised by various wavelet bases and then a set of the denoised images are obtained. Next, the local character is computed in time domain of every denoised image. The variance, the mean and the gradient are selected to explain the local character of the denoised images. When the character is bigger, the recovery effect of the image is better, the pixel is picked up as a clear pixel to the final denoised image. Third, these multiple denoised images are fused into a final denoised image based on the local character. Experiments show that the simple fusion improves the visual quality of the recovery image.
Page(s):
725-729
DOI:
DOI not available
Published:
Journal: Journal of Theoretical and Applied Information Technology, Volume: 46, Issue: 2, Year: 2012