基于同名点追踪的空间相机成像拼接配准模型

分析了已有图像配准算法在遥感图像拼接配准方面面临的问题。根据空间相机TDICCD交错拼接的成像特点,提出了一种基于同名点轨迹追踪的成像拼接配准模型。通过建立辅助空间坐标系下的中心投影共线方程,将像点、摄像中心、景物点建立严谨的数学关系,可精确实现对同名像点在像面上的轨迹追踪。结合TDICCD在像面上的位置约束,计算图像上同名像对的纵向偏差像元数和横向偏差像元数。最后结合相机在轨所成条带图像和卫星辅助数据进行分析,选取多组像点进行配准,同名像对配准误差小于1 pixel,经验证模型算法可行。相比传统的遥感图像配准算法,该方法不需要已知图像内容,为一种严格的几何意义上的配准,具有很强的适应性和预测性,已应用在型号相机的地面复算,易移植应用于其他类型空间相机的图像配准与拼接。

Registration model based on homologous points tracking of space camera assembly imaging

The defect of the existing imaging registration algorithms which are applied in remote sensing image assembling and registration was analyzed. According to the imaging characteristic of interleaving assembly for space camera, a registration model based on homologous points tracking for assembly imaging was proposed. The collinearity equation of perspective projection under assistant space coordinate was established. Then, the strict mathematical realationship of image point, camera imaging center and ground point was established. The locus of homologous points at image plane could be tracked accurately. Combined with the restrict of TDICCD on image plane, the number of pixels in direction of longitudinal and transverse displacement of pair of homologous points was calculated. Finally, according to the analysis based on image bands camera imaging on the track and satellite assistant datas, multiple of image points were selected to be registrated. The error of homologous points registration was less than 1 pixel. The feasibility of registration algorithms was proved. Compared with traditional registration algorithms for remote sensing image, this method is a strict registration with meaning of geometry, and there is no need to know the content of image. The method applied to the checking and calculating on correlative camera, has better adaptability and predictability. The model is easy to be applied to the imaging registration and assembling for others kinds of space camera.