大面阵航摄影像重建高效像素插值算法研究

A high efficiency pixel interpolation algorithm for large area-based aerial image reconstruction

大面阵或超大面阵数字航摄影像往往由多个子影像组成,且数据量较大,而最终合成影像的每一个像素都是从这些子影像内插而得,所以影像复合处理过程中涉及的几何形变等图像操作就会变得十分耗时,严重影响大面阵影像的重建效率,并影响相机的使用效能。文中对此提出一种快速高阶多项式图像插值方法,在尽可能不引入像素插值误差而损失影像自身固有几何量测精度的前提下,大幅度降低数据处理的计算复杂度,满足大幅面数字航摄影像重建对精度与效率的双重要求。

Because large or super large area-based aerial photography is often composed of multiple sub-images,the amount of data is very huge.In this way,the image warping and other pixel operation involved in the image preprocessing will become very time-consuming,and so the efficiency of composed large image reconstruction determinates the camera's performance.In this paper,a fast high-order polynomial image interpolation method is proposed,which can greatly reduce the computational complexity of data processing withoutthe geometric measurement accuracy of the image itself being lost much more or less,and then meet the requirements of efficiency of large-scale digital aerial image reconstruction.