Abstract:

Images captured in foggy weather conditions suffers from inevitable problems such as low contrast, blurriness and less visibility. In many computer and vision applications like surveillance, object detection, object tracking and navigation, these low-quality images cannot be used and it requires improvement in the image quality. Different algorithms have been proposed in this direction and to upgrade the quality of a foggy image. Most of the existing methods gives good quality image but with high time complexity. In this paper, a novel and effective method is proposed to remove fog from a single image. The proposed method is based on principal component analysis and modified dark channel prior. In proposed algorithm, foggy image is pre-processed using principal component analysis. This pre-processed image is further enhanced using fast global smoothening filter. Time complexity of the proposed method is much less as compared to the various existing methods and at the same time, quality is also maintained. Also, proposed algorithm does not require a large data set and specific hardware. To see the effectiveness of the proposed technique, both qualitative and quantitative analysis has been done on synthetic data set as well as on natural dataset.

Objective:

The objective of this project is to develop a fast and effective vision enhancement method for single foggy images, leveraging advanced dehazing techniques. The method aims to restore visibility, improve image clarity, and enhance object recognition by accurately estimating atmospheric light, refining transmission maps, and efficiently removing haze. The approach will focus on optimizing processing speed while maintaining high-quality image restoration, making it suitable for real-time applications in various fields such as autonomous driving, surveillance, and outdoor photography.