遥感卫星图像几何粗校正的数据并行方法研究

主要研究星上遥感图像的实时几何粗校正问题．卫星遥感图像现在一般都大到上万个像素行和列，采用传统的单个处理器的串行方式在星上进行实时处理是难以满足应用要求的．提出了一种在一维PE阵列的SIMD计算机上采用基于处理元阵列平移的数据并行校正方法，并根据NASA的LANDSAT-1卫星的有关的参数，对该方法进行了详细讨论，给出了具体的实现方法．通过对复杂性和加速比的讨论，表明该方法在性能上比采用单个处理器的串行方法提高了N倍．

A Data Parallel Algorithm on Geometric Correction of Remote-Sensing Satellite Images

Discussed in this paper is a real-time algorithm on geometric correction of a remote-sensing image on satellite. Nowadays,a remote-sensing image is more than 10 thousands of rows and columns larger and the traditional methods of sequential computing on a single processor can not meet the needs of real time processing on satellite. Proposed in the paper is a data parallel algorithm on one dimensional PE array of SIMD computer based on the shift of PE array,and under the parameters of NASA LANDSAT-1 the algorithm is discussed in detail,and the way of implementation is given. By discussing on complexity and speed-up of the algorithm,it is clear that this algorithm is N times higher in efficiency than the sequential algorithm.