摘要
遥感图像在数据采集和传输过程中受到各种噪声干扰,影响了图像本身的质量,不利于遥感图像的应用。本文利用最邻近算法(KNN算法),在基于图像处理器(Graphic Processing Unit,GPU)的统一计算设备架构(CUDA)架构下,实现了对遥感图像进行了去噪声的并行处理,并与在CPU下的串行处理做了对比,处理速度平均提高近3 000倍。遥感图像的其他处理也可以基于GPU实现。
The remote sensing data are influenced by many kinds of noises during the data collecting and transmitting,which is bound to decrease the quality of the image and isn＇ t contributing to the application of the remote sensing technique. In this paper,the KNN（ K-Nearest Neighbor）algorithm is employed to wipe off the noise of remote sensing image with parallel process based on the CUDA （Compute Unified Device Architecture）architecture of GPU （Graphic Pro- cessing Unit）. Compared to the serial processing on the CPU, the average processing speed is enhanced to nearly 3000 times. The other processing of the remote sensing image can also be completed based on the GPU techniques.
出处
《城市勘测》
2010年第2期96-98,共3页
Urban Geotechnical Investigation & Surveying
关键词
遥感图像去噪
GPU
KNN
并行处理
remote sensing image
GPU（ Graphic Processing Unit）
KNN（ K-Nearest Neighbor）
parallel processing
