Detection and Classification of Different Cloud Images for Meteorology Alert System

Abstract

Sky–cloud pictures procured from ground-based sky cameras are by and large discovered using a fisheye point of convergence with a wide field of view. In any case, the sky shows a colossal exceptional range similar to luminance, past what a standard camera can get. It is as such difficult to get the nuances of an entire scene with a standard camera in a lone shot. A great part of the time, the circumsolar region is overexposed, and the areas near the horizon are underexposed. This renders cloud division for such pictures irksome. In this paper, we propose HDR Cloud Seg – a practical system for cloud division using high-dynamic-go (HDR) imaging subject to multi presentation blend. We depict the HDR picture age method and release another database to the system for benchmarking. Our proposed approaches is the principle using HDR splendor maps for cloud division and achieves great results and besides mastermind the cloud types by using KNN portrayal.