【摘要】 高光谱遥感图像具有超多波段、光谱分辨率高、信息量丰富等优点,但同时也给异常探测的实时处理带来了重大考验。基于Cholesky分解的高光谱实时异常探测算法很好地解决了实时性问题,而图形处理器(GPU)的并行优化设计则更高效。实验结果表明:提出的优化设计在保证探测精度的同时,进一步提升了计算效率,算法加速比最高达到3. 14倍,说明基于GPU的并行优化算法能够较好地满足高光谱遥感图像实时处理的应用需求。更多还原

【Abstract】 Hyperspectral remote sesing images have the advantages of super multi-band,high spectral resolution,abundant amount of information,and so on. At the same time,it is a tough task for real-time processing of anomaly detection. The hyperspectral real-time anomaly detection algorithm based on Cholesky decomposition solves the real-time problem well,while the graphics processing unit( GPU) parallel optimization design is more efficient. The experimental result shows that the proposed optimal design significantly enhance the efficiency of computation,while assuring the precision of detection. This new parallel algorithm acceleration ratio reaches up to 3. 14 times,which shows that the GPU-based parallel optimization algorithm can meet the application requirements of real-time processing of hyperspectral remote sensing image well.更多还原