An expeditious cum efficient algorithm for salt-and-pepper noise removal and edge-detail preservation using cardinal spline interpolation

Keywords: Edge-preserving regularization Salt-and-pepper impulse noise Cardinal splines Expeditious algorithm Image de-noising Cardinal spline interpolation Edge preservation Two-phase noise removal a b s t r a c t This work proposes a faster and an efficient way to remove salt-and-pepper impulse noise and edge-preserving regularization of the henceforth obtained image. In this paper, we propose a two phase mechanism where the noisy pixels are identified and removed in the first phase. The detected noisy pixels in the first phase are involved in cardinal spline edge regularization process in the second phase. Promising results were found even for Noise levels as high as 95% with the proposed algorithm. The results were found to be much better than the previously proposed nonlinear filters or regularization methods both in terms of noise removal as well as edge regularization. Transmission errors, camera sensors having improper pixel elements, faulty memory locations and timing errors in analog-to-digital conversion [1] are the main reasons behind pixel corruption in digital images. However, only certain pixels are corrupted in such type of noise but the rest are noise-free. Denoising the image is a fundamental task since the subsequent output is used in other image processing operations such as, edge detection and regulari-zation, image segmentation, image zooming, and image compression object recognition. The two common types of impulse noise are the salt-and-pepper noise and the random-valued noise. In images corrupted by salt-and-pepper noise, the noisy pixels can take only the maximum and the minimum values in the dynamic range while images corrupted by random valued noise can take any value in between the minimum and maximum noise values in the same dynamic range. The goal of noise removal is to suppress the noise while preserving the other very fine details of the images. Different technical methods have been proposed to remove impulse noise. The median (MED) filter was once the most popular nonlinear filter. Its denoising power is elaborated in [2]. The computational efficiency of this filter can be seen in [3]. However, since every pixel in the image is replaced by the median value in its neighborhood, the median filter often removes desirable details i.e., vital information is lost in the image and moreover the image becomes blurred. The weighted median filter [4] and the center-weighted median filter [5] were proposed as replacements to median filter by giving more importance to some selected pixels (''decision based'') in the filtering …