Abstract: Based on the study of the principle of dual-tree complex wavelet transform, a cloud removal method was proposed by combing the frequency differences of cloud-mist and scenery.A hardware system for cloud and mist removal was designed and established.The system structure and development procedures were introduced.The design methods of the camera driver, video signal output program and man-machine interaction interfaces were described.Additional, the image acquisition and cloud-mist removal processing were implemented by applying the designed system, and the procession results were presented and analyzed.Results show that the entropy of this algorithm is 7.61 when the decomposition level number, the dividing number and the weight factors have the default values, proving that the proposed algorithm can achieve better results than the wavelet threshold method and the homomorphism filtering method.The designed system has a pleasant ability for cloud and mist removal as well as a strong adaptivity.