Fast radiometric correction method and implementation for area-array digital aerial image
With the development of computer hardware and software,especially the continuous progress of the programmable graphics processing unit(GPU)framework compute unified device architecture(CU-DA),the demand for developing image fast processing using CUDA has grown unprecedentedly.Hence,a CUDA-based fast radiometric correction method for all-color sub images of area-array digital aerial camera is proposed in this paper by analyzing the radiometric characteristics of DMZ Ⅱ,and the specific implementation steps and thread configuration methods are described.Moreover,the performance of the fast radiometric correction algorithm is tested by using the image data obtained by DMZ Ⅱ camera in a test area.The results show that,the proposed method can improve the computing efficiency by more than 13 times,and the logic is clear and the implementation is simple compared with the traditional central pro-cessing unit(CPU)serial programming method,so it can provide technical references for the develop-ment of related applications with similar image fast processing needs.