光学卫星在轨动态场景实时匹配方法及试验

On-orbit dynamic scene real-time matching method and experiment of optical satellite

针对目前在轨卫星图像动态范围偏窄、直方图集中、灰度层次不够丰富、暗场景图像细节分辨能力不强的问题,提出一种卫星在轨动态场景实时匹配方法。首先,研究云检测和基于直方图特性的大气程辐射预估方法,消除它们对场景高、低动态测量的影响,并结合测光相机与成像相机辐射响应关系的标定,通过测光相机最多2次拍摄地面场景,实现场景动态范围的实时测量;然后,针对地面场景动态范围通常超出相机动态的问题,设计并提出了基于高亮度和低亮度匹配的相机与场景动态范围匹配方案,同时给出了不同情况下相机在轨参数解算方法。最后,通过无人机飞行试验对匹配方法进行了试验验证,结果表明:利用该方法可根据实时拍摄的地面景物合理地设置相机积分级数和增益,实现相机与场景动态范围的最佳匹配,有效灰阶提升优于100%,信息熵提升优于40%。

On the basis of on-orbit dynamic scene real-time matching of optical satellite,a new method is proposed to solve some key problems in satellite imaging,including the narrow dynamic range,the coarctate gray scale distribution,the deficiency of gray level,and the lack of detail resolution in dark scenes.Firstly,the cloud detection and atmospheric radiation prediction methods are presented based on histogram characteristic,in order to diminish their influence to high and low dynamic metrics of scenes.Combined with the radiation relationship calibration between photometric camera and imaging camera and image motion compensate,the real-time measurement of the dynamic range of the scene is achieved by imaging twice(not more than)using the photometric camera.Then,aiming at the problem that the scene dynamic range usually exceed the camera dynamic range,a matching method between the camera and scene dynamic range is proposed based on the high-and-low luminance matching schemes.Meanwhile,the camera parameters'calculation method in different situations is presented.Finally,the proposed matching method is verified by experiments of unmanned aerial vehicle(UAV).The results indicate that the method can achieve the best matching dynamic range by setting suitable integrating numbers and gains of the camera according to the actual imaging dynamic scenes,which can improve the effective gray scale and the image entropy by 100%and 40%,respectively.