Abstract:

Aiming at the problem of image processing, we proposed a fast single image dehazing method based on feature fusion, and solved the problem of the block effect of dark channel method. By means of the dark channel prior based on K-means clustering, the coarsescale transmission was obtained. By analyzing the effect of haze on imaging, we extracted the saturation of hazy image itself as finescale transmission, which could reflect the change of scene depth effectively. Finally, we obtained accurate transmission via image fusion technique. Through the test of a variety of realscene images, the experiment results show that the method is simple and effective, and can get ideal dehazing result.

Key words: image dehazing, K-means clustering, multiscale, feature fusion