摘要
提出了一种在CUDA架构上实现基于Bresenham画线算法的图像快速旋转算法.该算法避免了复杂的数学运算,并融入Bresenham画线算法的优点,解决了大量的浮点的取整运算问题,并且减少了图片旋转过程中精度的损失,使得算法的精度和时间复杂度达到相互平衡,极大提升了旋转算法的性能.实验证明,在当前的硬件平台下,与传统旋转算法相比,该算法极大的提高了旋转算法的速度并减少精度的损失.
To accelerate the speed of image rotation and reduce the accuracy loss in the conversion process, a fast image rotation algorithm is provided based on CUDA architecture and Bresenham Line drawing algorithm for solving the rounding operation of floating point and reducing the accuracy loss in the rotation course. It can balance the accuracy and time complexity of the algorithm, which can achieve performance improvements of the rotation algorithm. The experiments show that compared with the traditional algorithm, the method has a greatl.y improved speed of the rotation algorithm and reduced accuracy loss in the current hardware platform.
出处
《大连交通大学学报》
CAS
2012年第3期63-66,共4页
Journal of Dalian Jiaotong University
关键词
CUDA
旋转
多线程
CPU计算量
CUDA
rotation
multi-threaded programming
CPU computation