Research On Key Technologies For Optimizing Tile Generation Of Remote Sensing Image Data

With the development of satellite technology and the improvement of sensor accuracy,the data volume of remote sensing images is increasing by the geometric progression.However,the existing tile generation technology is slow in processing large-scale remote sensing images,and the processing efficiency is low..Therefore,how to realize the rapid tiling of remote sensing image data is the key to break through the bottleneck of the visualization efficiency of remote sensing image data.This paper starts with image resampling interpolation and image cropping algorithm,and studies how to optimize the image generation efficiency of remote sensing image by optimizing the image interpolation algorithm and image cropping algorithm.Specifically carried out the following research work:(1)Comparison test multiple image interpolation algorithms—select the best interpolation algorithmImage resampling interpolation is a key step in the generation of tiles by remote sensing images.Most of the existing image interpolation algorithms used in software tools with tile generation function are nearest neighbor interpolation.Although the algorithm is small in calculation,it is processed.The image will have a certain degree of distortion,which affects the browsing experience of the tile map.In order to solve this problem,a more suitable interpolation algorithm is chosen to replace the nearest neighbor interpolation method.Through test comparison,a bilinear interpolation algorithm with better picture processing quality and moderate algorithm complexity is selected.(2)Improved resampling interpolation processing—introduction of GPU acceleration technologyWhen re-sampling the remote sensing image data,it is found that the process is repeated for the operation of the image pixels,and has parallelism.Based on this feature,the GPU parallel acceleration technology is introduced to optimize the interpolation process of the serial task originally performed in the CPU.Transferred to the GPU parallel task,effectively improve the calculation speed of the interpolation algorithm,and achieve fast resampling of remote sensing images.(3)Improve the tile generation speed of the cropping algorithm—multi-threaded task parallelBased on the cutting algorithm of GDAL,it is improved to make the algorithm segment the remote sensing image according to the rules of tile generation.On this basis,the multi-task division function is added to the algorithm,so that the algorithm can multi-thread parallel and speed up the cutting.Compared with the commercial software commonly used in the market,in addition to the processing speed has increased,more flexible.