A Simple Technique for Removing Snow from Images
Nishatha Nagarajan, Manohar Das · 2018
This paper presents a simple technique for removing snow from images that has applications in many areas of image processing, computer vision, and pattern recognition. The biggest challenge in this case is the variability in data. It is very important to differentiate patterns representing snow from other valid objects within an image. Since an image may contain a wide variety of patterns, a probabilistic method should be applied. In this paper, we present a snow detection and removal technique based on mixture modeling and compressive sensing. Mixture modeling is the primary tool used to detect and remove snow. During this process, however, some noise is induced in the processed images. Compressive Sensing Orthogonal Matching Pursuit algorithm is then applied to denoise the image. The final result shows an image cleared of snow, while still retaining the main objects and features of interest.