ABSTRACT:
In bad weather conditions like haze and dimness, the particles present in the environment dissipate episode light in various headings. Subsequently, picture taken under these conditions endures from lessened perceivability, absence of complexity, thus, it shows up lackluster. Picture dehazing technique endeavors to recoup a cloudiness free depiction of the given dim picture. In this paper we propose a technique that dehazes a given picture by looking at different yield patches with the first dim form and picking the best one. The correlation is performed by our proposed dehazed fix quality comparator in light of Convolutional Neural Network (CNN). To choose the best dehazed fix we utilize double inquiry. Quantitative and subjective assessments demonstrate that our strategy accomplishes great outcomes in the greater part of the cases, and are, on a normal, tantamount with cutting edge techniques.
BASE PAPER: Learning a Patch Quality Comparator for Single
