Image Dehazing using Improved Dark Channel Prior and Relativity of Gaussian

Abstract

Abstract Bad climatic weather conditions occur in our daily life are snow, sandstorm, mist, haze and fog etc. These various atmospheric particles affect the natural scene and the visibility will be tampered. Different dehazing methods are used to restore the visual perception of the image, increase the natural panoramic view and preserve the structural features of the image. Poor visibility becomes a significant issue for most outdoor vision applications. Image dehazing is a highlighted research area because of its real time applications in surveillance systems, driver assistance system especially for the people who resides in hilly areas where mist and haze is prominent. This paper presents an modified dark channel prior based image dehazing system with an enhanced and refined transmission map constructed by utilizing the RGB and YCrCb color system. Transmission map is constructed and estimated from combined airlight component obtained from the above said color models. In this work, we compute the Relativity of Guassian for RGB and YCbCr color space to produce guidance image. The both guidance images thus obtained are applied to guided filter for smoothening. Finally, hazeless image reconstruction is performed by utilizing the guidance images obtained from the previous step. In this proposed work, we focus on improving the transmission map by utilizing the concept of Relativity of Gaussian and Guided filter. We can find that the proposed method does not over enhance the images. Results analysis focuses on investigating the efficiency of proposed method by reconstructing the image without distortion, with no color artifacts and better aesthetic visual image