基于GPU和矩阵分块的增强植被指数计算

Calculation of Enhanced Vegetation Index Based on GPU and Matrix Partition

增强植被指数(enhanced vegetation index,EVI)算法是生态遥感领域的重要算法,被广泛应用在植被分类、作物长势检测和自然灾害监测等方面。而随着遥感影像获取能力的不断提升,在使用传统的EVI算法处理数据量较大的影像时会出现内存占用率较高且耗时较长的现象,给应用系统的快速响应带来了不便。针对这种情况,结合EVI算法特点,提出一种基于GPU和矩阵分块的EVI算法,充分利用CPU和GPU各自的优势提高算法执行效率。对于需要分块处理的影像数据,该计算方法根据影像数据大小、系统可用内存和GPU可用显存计算出分块数目,在CPU端将影像数据按照相同的规则进行分块,然后将分块后数据在GPU端进行EVI运算,最后将运算结果返回到CPU。实验结果表明,EVI算法经过改进后的运算速度得到了有效的提高,内存使用率得到有效的降低,验证了被改进算法的优越性。

Enhanced Vegetation Index（EVI）algorithm is an important algorithm in the field of ecological remote sensing,which is widely used in vegetation classification,crop phenology monitoring,natural disaster monitoring and so on.But with the continuous improvement of remote sensing image acquisition capabilities,traditional EVI algorithm requires higher memory usage and is time-consuming when used in processing large amount of data image.It leads to inconvenience for the quick response to operating system.To address this problem,considering the characteristics of EVI Algorithm,we propose an EVI algorithm based on GPU and matrix partition in this paper.For the image data which needs to be divided into blocks,this calculational method calculates the block number according to the image data size,system available memory and GPU available memory,divides the image data into blocks according to the same rule in the CPU,then executes the EVI operation in the GPU and returns the final result of the operation to the CPU at the end.Experimental results show that the computational speed of EVI algorithm has been effectively improved,and memory usage has been effectively reduced.Thus,the experimental results verify the superiority of the algorithm improved.