The development of supersonic especially hypersonic aircraft is an important reflection of the national defense strength. Therefore, the developed countries have invested a large amount of materials and financial resources to carry out the research on supersonic aircraft. While supersonic aircraft flying into the atmosphere with high speed, drastic interaction occurs between the vehicles’ optical hood and the high-speed airflow, thereby causing the aero-optical effects. As a result, the images received by the imaging system on the aircraft are blurring, distortion or shifting, which seriously affect the capability of aircraft for target detecting, recognition, tracking and precise attacking. Consequently, it is urgently need to study the image restoration theory and methods caused by the aero-optical effects. Among all, the blurring image restoration is a pressing difficult problem. Based on the supersonic aircraft, the paper carries out researching of restoration algorithms and its application of blurring image. The restoration of blurring image is the "inverse problem" in mathematics physics which is tough in theory and practice because of ill-posedness. Based on analyzing the research results of scholars at home and abroad in depth, the deblurring problem is studied in the paper. Firstly, aiming at overcoming the shortcomings of the Richardson-Lucy (RL) algorithm, that is, ringing artifacts and noise amplification, a new image deblurring algorithm is proposed by introducing Gain-Map, which avoids the spreading of ringing artifacts and further amplification of the image noise during iteration. Secondly, to solve the ill-posedness of the deblurring problem, Bayesian deblurring algorithm based on the Non-Local Range Markov Random Field (NLR-MRF) prior constraint is proposed, which can capture the local image details while acquiring structural information among adjacent blocks . Finally, to avoid the defects of the maximum a posteriori (MAP), a new deblurring algorithm based on minimum mean square error (MMSE) estimation is proposed. The new algorithm performs image deblurring and noise estimation in a unified framework which prevents additional noise estimation step or specifies the regularization parameter beforehand. In addition, through algorithm transplantation and optimization, the deblurring algorithm has been verified embedded prototype systems based on DM6467 from TI company. The main work and contributions are as follows: (1) A novel image d...
修改评论