Research On Detection And Compensation Technology Of Forward Image Motion In Aerial Photography Based On Image Restoration

In this dissertation, the technology of the forward image motion's detection and its compensation in aerial photography is discussed in detail, and the evaluation of the restored image blurred by image motion is introduced. At first, the technology of aerial imaging and the development of domestic and overseas aerial reconnaissance camera are stated, the effect on the aerial camera's imaging quality of the image motion, the necessity of the image motion compensation and its development are discussed, and the aerial camera's image motion are classified. It is illustrated that the forward image motion is the key one in all the image motions, its mathematical model is established and its discretizing is completed. The mechanism and their spectrum analysis of the forward image motion blurring and the defocusing blurring are given, and the spectrum transform of the blurred image is utilized to measure the quantity and the direction of the image motion. An effective rule to judge the forward image motion or the defocusing is constructed by analyzing their effect on the image gray. Then the classification, application and development of the digital image's restoration are discussed. For the restoration of the motion blurred image, several typical methods are studied including the Inverse Filter, Weiner Filter, Difference and the multi-frame interfusion restoration, analyzing their application and limitation. Subsequently a restoration method based on wavelet transform is presented in accordance with its mathematic characteristic. The development of the image quality evaluation, the traditional and the HVS based new ways are discussed. According to the analysis, the subjective evaluation and the objective evaluation based on parameters are combined to estimate the restored image's quality. The condition of forming the aerial forward image motion is simulated to imaging in lab. The measurement of image motion and the restoration to the preprocessed motion blurred image are carried out, the precision is less than one pixel, and it is deduced that the wavelet-based restoration is superior to the Difference and the Weiner Filter restoration.