IMPROVED TWO-PHASE TECHNIQUE FOR REMOVAL OF VERY HIGH DENSITY SALT AND PEPPER NOISE IN GREY SCALE IMAGES
Gagandeep Kaur1, Kulwinder Singh2*
1&2 Bhai Maha Singh College of Engineering, Sri Muktsar Sahib
Corresponding Author Email: monga_kulwinder@rediffmail.com
ABSTRACT: In this digital era, huge amount of digital data is transferred from place to another due to vast digital technology. The transfer of data and images from one point to another plays a vital role in this digital world. One of the common examples of digital data is image. During transferring of image, it may lose its quality and it is also sensitive to noise which degrades the quality of the image or destroy its edges. To overcome all these problems image processing is used. It is nothing but the processing of image using mathematical operators in which input is an image and output is desired characteristics of the image. In image processing, non-linear filters plays a vital role in removal of salt and pepper noise or impulse noise as linear filter fails to do so. In literature several non-linear filters were proposed to get better denoised image along edge preservation. But at very high noise density the existing non-linear filter either fails to preserve edges or fails to get better denoised image at noise density as high as 99%. In this thesis, a modified two-stage algorithm is proposed which is the fusion of best existing non-linear filtering techniques, retain the denoised image as much as possible. The proposed algorithm is tested for different grayscale images. The qualitative and quantitative results are examined by performance metrics Peak Signal to Noise Ratio (PSNR), Mean Square Error (MSE), Image Enhancement Factor (IEF) and Structural Similarity Index Measure (SSIM). The worth full results of the proposed algorithm are compared with several existing non-linear filtering techniques and it has found that the proposed algorithm gives best results in terms of different performance parameters as compared to different non-linear existing filtering techniques.
Key Words: Salt and Pepper Noise (SPN), Image Processing, Image Denoising, Peak Signal to Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Image Enhancement Factor (IEF)