Speaker
Description
This dissertation is about creating a non- redundant, effective and low- complexity denoising method. Denoising an image is involves removing noise from an image to keep the original elements of the image and remove the unwanted additions. Transform- based denoising depends on the transform used in the denoising method. Recent works focus more oin improving the performance of the denoising, with ignoring the complexity. This work studiesy the complexity of well-known transforms that that capture texture in images sufficiently, known as Contourlet transform (CT) and non-uniform directional filter bank (NUDFB). Complex wavelet transform (CWT), CT and, non-subsampled Contourlet transform are examples of the base of denoising methods for in the majority of current works. All these transforms perform well but they are complex and involve redundancyt transforms. Noise that applied to images in this work is Gaussian noise and all images used are greayscale images;. The proposed work here used the Linear Minimum Mean Square Error (LMMSE) method was used for denoising, and with NUDFB as the base transform. The proposed method decomposes the image using NUDFB then recognizes the coefficients to the step of denoising. LMMSE compares the coefficients in different resolutions to predict noise and remove noisethem. For measuring work accuracy, Peak Signal to Noise Ratio (PSNR) and Structural Similarity (SSIM) are used. The resultinged images from the proposed denoising method shows improvement in the results. The PSNR of the proposed method is higher than thresholding using NUDFB, about 1db. When comparing the proposed method with wavelet transform thresholding and CT, proposed methodit has higher values than in CT and WT, especially in a high noise ratio. In images that contain directional information, such as fingerprint images, the proposed method has the highest SSIM. The pProposed denoising method creates a way of denoising images using less fewer requirements because of the low complexity. It is also has high better results that can out-perform over heavy methods that uses soft thresholding methods.
