Image Denoising Based On Lifting Wavelet Transform

In real life, the images will inevitably be affected by noise during the collection and transmission process, this will result in image quality. People have been looking for denoising method which can effectively reduce the noise while retaining a good image edge information. Compared with the Fourier transform, wavelet transform in the time domain and frequency domain has a good localization of the nature and characteristics of multi-resolution analysis, and therefore it is widely used in the image noise reduction processing. However, in the actual image denoising application process, the traditional wavelet transform tends to loss image detail easily, there operation was slow, and some disadvantages such as storage space requirements. In contrast, as the second-generation wavelet, with its unique algorithm structure, lower computational complexity and easy to achieve integer-to-integer transform, etc., lifting wavelet has obtained more attention and favor by experts and scholars, and the application on the lifting wavelet has become a new hot spot in areas of wavelet research and applications.Firstly, the paper describes the development of image denoising and the traditional wavelet transform basic theory, image edge detection theory is given, the principle of wavelet denoising and the frequently-used types of wavelet denoising method are introduced, threshold functions selection and threshold determination in threshold denoising method are described in detail, while various methods are compared together by numerical simulation.Secondly, the paper introduces lifting scheme of the second generation wavelet, gives the process of traditional Matlab algorithms by using lifting scheme, and analyses features of lifting wavelet transform theory and lifting wavelet transform, and applies it to edge detection. Based on lifting wavelet and canny operator, edge detection method is proposed, the experiment results illustrate the anti-noise of the algorithm proposed.Finally, image denoising method of edge enhancement based on lifting wavelet is proposed, directed to the traditional denoising technique (such as wiener time-domain filtering)because of too much loss of image edges and details. Considering improving the image edge and texture detail using edge detection and image fusion method before image denoising, and then we can decompose image by lifting wavelet transform, deal with the high-frequency using adaptive image denoising at last. On account of the image edge and texture details are enhanced before denoising, that is, their corresponding wavelet coefficients are amplified, thus reducing the shrinkage possibility of these wavelet coefficients in the threshold value denoising process. The simulation shows the effectiveness of the algorithm, Except for image denoising, this method can preserve image edges information better as well.