An adaptive grayscale image de-noising technique by fuzzy inference system

Noise is an issue for image in terms of corruption. Some noises can be automatically added with the image while capturing the image. Camera sensors and lighting factors are responsible for those. Noises that can be added with the image is known as additive noise. To get a perfect image noise reduction is very important. Edge detection, image enhancing, object detection etc. are directly related with noise reduction. Noises need to be redacted before image processing. There are many well established noise reduction techniques available. Linear noise reducing algorithms are not efficient to reduce impulse noise as they are mainly focused on smoothing that makes the edges of an image blur. On the contrary, nonlinear algorithms do a great job for reducing impulse noises. In this paper, we have made a proposal which is an efficient algorithm for digital grayscale image de-noising using fuzzy logic. Our proposed algorithm uses the Fuzzy Inference System (FIS) for the membership value calculation. The noise affected pixels are determined by the FIS from the 3 × 3 neighborhood. Our method works very fine with impulse noises. The proposed algorithm returns almost the original noise free image which has a significant difference in the Peak Signal to Noise Ratio (PNSR) compared to the existing Median and Average filtering technique.