An Efficient Salt and Pepper Noise Reduction Approach for Video(s) Using Optimized Filter Approach

Abstract

In today's digital world, most information is shared through images or videos. Images and videos are more infor-mative than textual data. During the image or video capturing process, addition of noise is one of the major problems. The presence of noise in captured data causes results to be misleading, particularly in computer vision and image processing tasks. As a result, preprocessing is required before tackling tasks like edge detection, object detection, object recognition, salient object detection, video summarization, and so on. Gaussian, Salt & Pepper, Poison, and Speckle noise are the most common types of noise which can affect the video. In this article, the author(s) presented an efficient noise reduction approach for reducing the salt and pepper noise present in the image or video. The performance of the proposed approach is compared with traditional and widely used filters for noise reduction, such as mean, mode, and median. The experimental analysis of the proposed approach is done on the dataset where the videos are captured using the camera as well as from the other resources such as Internet. Noisy videos are prepared by mixing the Salt & Pepper noise with different noise densities (levels) in the video dataset. The experimental results show that the proposed approach outperforms the traditional approaches not only in terms of noise reduction but also in preserving the details in the video.