Human-computer interactive system development for cloud removal from single images

Abstract

Remote sensing imaging is disturbed by clouds frequently and produces images with low contrast and poor resolution. Algorithms removing clouds from single remote sensing images are expected to improve their qualities and aroused worldwide interests. Compared with multi-image methods, cloud removal algorithms from single images reduce the demands for weather and improve the image collecting flexibility. In this paper, a human-computer interactive system for cloud removal from single images was developed according to the proposed cloud removal algorithm. Firstly, the distributions of clouds and sceneries in the contaminated remote sensing images were discussed. Features of the dual tree complex wavelet transform (DTCWT) to split image frequencies were investigated after the principle of DTCWT was analyzed. By combining the frequency features of DTCWT sub-bands and the frequency distributions of the remote sensing images, the cloud removal algorithm based on DTCWT was proposed, and its procedures were described completed. Then, a human-computer interactive system for cloud removal was developed in order to improve the cloud removal processing conveniences and achieve the best performance. The development schemes were introduced. The system structures were illustrated. The design methods of the camera driver and video signal output program were described. Communication streams between modules and human-computer interaction interfaces were designed further according to the cloud removal algorithm based on DTCWT. The system has a very good flexibility which permits the input of most parameters from users and is expect to achieve a best performance. Finally, the device of the system was completed and tested. Image acquisition and cloud removal processing experiments were implemented by applying the human-computer interactive system. Cloud removal effects by adjusting parameters through the human-computer interaction interfaces were compared and evaluated. It proved that the designed system has a pleasant human-computer interaction ability and satisfactory cloud removal performance.