Color Attenuation Prior Based Single Image Dehazing

Abstract

In today's technologies, digital images of the real world play a crucial role. Any mode of transportation can employ digital imagery to control traffic. Clearly stated Images are essential for traffic monitoring, but our environment is full of dust particles, fog, and other disturbances depending on the weather. Haze is the aggregate name for these disturbances that scatter ambient light and subsequently affect the clarity of the images that are taken. The images that we take often have haze, thus we refer to them as hazy images.The collected images must have the haze removed in order to be monitored. Dehazing, often known as the process of removing haze, can be carried out in a variety of methods. In this research, we employ the Haze lines Prior method of estimating atmospheric light along with the Color Attenuation Prior method for dehazing. The databases O-haze, I-haze,FRIDA and some hazy real world images without ground truth images are used to validate the proposed method. By analyzing PSNR and SSIM values on a pair of ground truth and hazy images, the algorithm's effectiveness is evaluated.The performance metrics of the suggested algorithm produced outcomes that were superior to those of the existing methods.