LAUSR

The prognosis of disease detection is highly dependent on the input obtained from primary-sources. To substantially increase the quality of image, a procedural denoising algorithm is required. In that aspect, a novel denoising-algorithm based on adaptive conductance-function in the diffusion-filter is proposed. Here, the noise-contaminated image is filtered and pixel-corrected by incorporating local-difference and thresholding methods. Later on, bidimensional empirical mode decomposition is employed to decompose the image into respective mode-functions, followed by optimization procedure based on proposed objective function to optimize the threshold-gradient and iteration parameters of the modified anisotropic diffusion process. This process ensures not only an improvement in denoising performance, but also enhances the visual quality of the noisy-image. Finally, the algorithm has been verified with the inclusion of hardware implementation using Raspberry-Pi-5 (RPi-5). This algorithm uses mean-square-error(MSE), peak-signal-to-noise-ratio (PSNR), structural-similarity-index (SSIM), and feature-similarity-index (FSIM) as performance indicators to be compared with different denoising-techniques applied on various medical images, and achieves an average decrement of 81% with 3.70% deviation in MSE, 50% increase and 0.89% deviation in PSNR, 770% increment and 5.22% deviation in SSIM, and an increment of 57% and 1.31% deviation in FSIM when competed with algorithms implemented on a personal computer and RPi-5 hardware.