Research On Global Motion Vector Estimation For Video Stabilization In Flight Measurement And Control

Optical measurement plays an extremely important role in spacecraft TT&C technology.However,environmental vibration or irregular movement of platform will cause optical camera jitter,which produces video jitters.Video jitter will affect completion of TT&C tasks.Thus,the image stabilization is essentially researched to eliminate or reduce jitters.The platform of existing optical TT&C equipments have two kinds,namely fixed platforms and airborne platforms.Fixed platform cameras have jitters caused by the random jitter of joystick,environmental noise and other factors;airborne platform cameras will also generate jitters,which due to irregular movement of platform movement.In order to eliminate the jitter caused by those jitters and movement,this paper deeply analyzed the special characteristics of video jitter in aircraft TT&C and studied the key techniques of global motion vector estimation,which can be concluded the following aspects:1.A fast and accurate global motion vector estimation method based on sub-pixel phase correlation and multiple sub-blocks collaboration is proposedFor fixed platform camera,a fast and accurate global motion vector estimation method is proposed based on the analysis of video jitters characteristics.Firstly,the sub-pixel sampling techniques and fast sub-pixel phase correlation are combined to calculate the motion vectors of individual sub-blocks.Then,multiple sub-blocks collaboration method is used to remove the interference of local moving target.Experimental results show that the method has the advantages of high accuracy,low complexity and strong environmental adaptability.2.A fast global motion vector estimation method with the combination of prioriknowledge and SIFT method are proposed which has the rotation processing facultyThe motion of airborne platform camera includes intentional scanning of cameras and intentional motion of platform.Jitters of video include both translation and rotation.The existing algorithms are not able to take full advantage of the characteristics for jitter video captured by airborne platform,which makes calculation accuracy low and real-time poor.Thus,a new method for global motion vector estimation is proposed.Firstly,an approximate video motion model is proposed based on the analysis of airborne platform video motion characteristics.Secondly,the prior knowledge of airborne digital image stabilization is summed up.Thirdly,global motion vector estimation method is proposed based on the combination of prior knowledge and SIFT features.The method uses the prior knowledge to reduce SIFT feature points number and simplify feature points descriptions,which reduced the computational complexity.The method eliminate the mismatching feature points couples with SIFT,which increases the accuracy.The experimental results show that this method not only has the accuracy and anti-jamming ability of SIFT algorithm,but also greatly reduces the computational complexity.