AN EFFECIENT DENSELY-CONNECTED PYRAMID DEHAZING-NETWORK

Abstract

Imaze quality may degrade by scattering particles in atmospheric particles for ex. fog and smoke. Due to this image can have diminished contrast, altered colour, and blured effect .This makes it hard to be recognized by human eye and computer vision systems . Hence Image dehazing is necessary which is a growing technique for dealing with the deterioration of such photographs. This research proposes dehazing method based on densly connected pyramid structure, end-to-end single picture dehazing system. First we take the input image and combine it with dark channel to make new input which is then further put into generator made of densly connected pyramid structure. The output of this geberator is then given to discriminator along with dataset to find discriminator score. This way the dehazed image is achieved .All components are discussed in detail further in this report.