高精度机载可见光图像定位方法

针对现有技术不能解决机载可见光图像快速自动定位的问题,提出了一种定位精度高、处理速度快、可扩展性好的机载可见光图像定位方法.该方法以传感器共线成像模型为中心,将实时图像和传感器成像参数送入共线模型进行正射校正,校正后的图像与相应的基准图像进行图像配准,使用配准同名点信息更新传感器成像参数,最后再次通过共线模型获取实时图地理坐标定位信息.这种通过图像配准再计算成像模型的间接定位方法,不仅减少了计算量,而且使自动配准算法稳定,较传统的直接定位方法或图像配准方法有突出的优势.此外,该方法具有严格的误差传递计算公式,能计算定位结果的误差.仿真实验表明即使在大倾斜角条件下,该方法也能获得优于100 m的定位精度.

A method for high accuracy airborne optical image geolocation

For the unsolved technology problem of fast automatic geolocation of airborne optical images,a high accurate,high efficient and good scalable method is proposed for geolocation of airborne remote sens-ing optical images is proposed. The method is centered on sensor collinearity equations. The real-time im-age and the sensor imaging parameters will be put into the collinearity model for geometric rectification. Then the rectified image is registered with corrresponding reference image. The matching points are used to update the sensor imaging parameters. Finally,the updated parameters and real-time image are put again into the sensor collinearity model for acquiring geometric location information. This indirect geolocation method,which is based on image registration and imaging model,not only reduces the computation,but also makes the automatic registration stable. It has a prominent advantage over traditional direct or image regis-tration methods. In addition,a strict error propagation formula is capable of computing the geolocation ac-curacy range. Simulation results show that even under the highly-oblique condition,a geolocation accuracy better than one hundred meters can be obtained with the proposed method.