Abstract Views :167 |
PDF Views:2
Authors
Affiliations
1 Department of Computer Science and Engineering, Supreme Knowledge Foundation Group of Institutions, Mankundu, Hooghly, IN
2 Department of Computer Science and
Engineering, Supreme Knowledge Foundation Group of Institutions, Mankundu, Hooghly, IN
3 Department of Computer Science and Engineering, Netaji Subhash Engineering College, Techno City Garia, Kolkata, IN
Source
Digital Image Processing, Vol 7, No 8 (2015), Pagination: 242-246
Abstract
In this paper, it has been intended to detect the random valued noise and then remove it with the approximation of neighboring pixels. The detection process comprises of two parts. One for the border detection and other for the detection of the rest of the image. In detection process median is computed taking fixed 5×5 window. A pre-defined threshold value is set for the detection of corrupted and un-corrupted pixels. In removal process row wise and column wise matrix operations are separately performed on two distinct images. The output of the above two operations are merged together to get a new matrix. Then conditional mean operation is performed to replace the noisy pixels. Lastly border removal is done and the overall image is further smoothed by unconditional mean operation. Experimental result shows that the proposed filter outperform other filters in respect of performance at noise density as high as 65%.
Keywords
Random Valued Impulse Noise, Mean Filter, Row Wise Operation, Column Wise Operation, Merging, PSNR.
