Modified Box Filter Design and Noise Analysis on Two-Dimensional Images

Box filters have been used to improve several potentially challenging image processing procedures. The Modified Box Filter has been designed to denoise the images in this work. The box filter algorithm is used for image enhancement on 2D images and various 2D images are tested in this process with different noises like Gaussian, Salt and Pepper, Poisson, and Speckle noise. The results from the designed filter with improved parameters such as peak signal-to-noise ratio (PSNR), normalized correlation (NC), structural similarity index (SSIM), mean square error (MSE), and multiscale structural similarity index (MS-SSIM) indicates improved image quality in 2D images. The proposed Modified Box Filter algorithm performs better against Poisson noise when compared to other noises which are evident from simulation results. It provides PSNR = 37 dB, SSIM = 87%, NC = 98%, MSE = 0.001%, MS-SSIM = 85%.When compared to existing methods, the proposed technique produces better results, making it a good choice for 2D image improvement.