Comparison of Modern Denoising Algorithms

Integrity of edge and detail information associated with the original images play an important role in applications. Images acquired from sensors, transmission errors and lossy compression contains noise and it is necessary to apply an efficient denoising technique to compensate for such data corruption. Image denoising still remains a challenge for researchers, since noise removal introduces artifacts and causes blurring in images. The median filter and specialized median filters are most popular for removing salt and pepper noise however when the noise level is high some details and edges of the original image are smeared by the filter. Decision based Tolerance based Selective Arithmetic Mean Filtering Technique (TSAMFT) algorithm works very well but if the noise density is high, then the image recovered using TSAMF is not good. Improved Tolerance based Selective Arithmetic Mean Filtering Technique (ITSAMFT) provides best results for removing salt and pepper noise even for higher noise density levels and it preserves the best edges and fine details. Comparison of these algorithms provides a suitable basis for separating noisy signal from the image signal. This paper presents a performance evaluation of Level-1 and Level-2 ITSAMFT, TSAMFT and Median Filtering algorithms in the detection and removal of Salt and Pepper Noise. The simulation results shows that the Level-2 ITSAMFT is superior over the Median Filter and TSAMFT in maintaining high peak signal to noise ratio (PSNR), correlation (COR) , image enhancement factor (IEF) and is more powerful algorithm in removing the heavy salt and pepper noise.