Publication

Abstract : A crucial step in multimedia processing is image and video denoising, which aims to eliminate undesirable noise while keeping significant image properties. This research investigates the use of four distinct multiscale transforms for picture denoising: ridgelets, shearlets, curvelets, and wavelets. To capturing and displaying visual structures at different scales and orientations, each transform offers special features. Through studies on artificial and real-world noisy images, examine each method’s denoising performance and compare its capacity to reduce noise while maintaining image information. The outcomes demonstrate the advantages and disadvantages of each strategy and shed light on how well they work with various kinds of noise and image content. This study investigates the effects of four different multiscale transforms (ridgelets, shearlets, curvelets, and wavelets) on the Signal-to-Noise Ratio (SNR) in addition to examining their denoising capabilities. SNR values are important measurements for assessing how well denoising works since they offer measurable information on how to strike a balance between noise reduction and detail preservation in the image. Through extensive experiments on synthetic and real-world noisy photos, this study evaluates the effectiveness of every denoising technique over a range of noise kinds and image contents. The abstract also emphasizes the distinct qualities that every multiscale transform offers in terms of capturing and expressing visual structures at different scales and orientations. By means of a thorough examination, the research reveals the benefits and drawbacks of every technique, providing significant understanding regarding their appropriateness for various denoising situations.