Structural Similarity Index based Evaluation of Dark Channel Prior Algorithm

Image dehazing and image defogging work towards the same goal, which is to reconstruct hazy or foggy images by eliminating noise. Among the many dehazing and defogging algorithms available, Dark Channel Prior (DCP) is among the most effective algorithm. DCP employs the dark pixels from the noisy image to perform image reconstruction using the standard four-step process. In this paper, we implement DCP single image dehazing for license plate images. Specifically, the transmission map refinement was implemented using the guided filter. The implemented model performs the DCP implementation along with its five important parameter evaluations and analyzes the performance. The paper concludes with a comparative analysis for five parameters that are patch size, epsilon value, window radius, value of $\omega$ and finally pixel values.