基于薄板样条插值算法的巨幅影像分块并行处理

Huge image block parallel processing based on thin plate spline interpolation algorithm

薄板样条函数是空间插值中的一种重要方法。对于巨幅影像数据使用薄板样条函数进行空间插值时,可能会出现运行时间太长,以及计算机内存空间不足或程序运行无响应的问题。针对这些问题,根据薄板样条函数光滑、连续的特点,基于GDAL开源函数库,提出对巨幅影像数据的分块读取,在块内利用并行技术求解线性方程组,确定薄板样条函数,最后进行空间插值的方法。结果表明,该方法可以有效的解决这些问题。

Thin plate spline is an important algorithm of spatial interpolation. For huge image data when using thin plate spline interpolation, the problem of running too long and insufficient computer memory space or program runs no response may occur.. To solve the problem, according to the smooth, continuous feature of thin plate spline, and based on GDAL open source library, proposes a method, in which the huge image data is divided into blocks, parallel technology is used to solve the linear equations in the block, and then interpolates with thin plate spline. The results indicate that it is an effective method to solve the problem.