CAJ | 학술논문

运动估计模块是数字图像/视频处理中最重要的模块之一，它的处理技术直接决定了图像/视频处理的效果，快速有效的运动估计算法是现代数字图像/视频处理所急需的。本文利用了一种并行的编程思想，在不影响算法质量的前提下，加速了普遍使用的一些运动估计算法。实验结果表明加速比可以达到8倍左右。
운동고계모괴시수자도상/시빈처리중최중요적모괴지일，타적처리기술직접결정료도상/시빈처리적효과，쾌속유효적운동고계산법시현대수자도상/시빈처리소급수적。본문이용료일충병행적편정사상，재불영향산법질량적전제하，가속료보편사용적일사운동고계산법。실험결과표명가속비가이체도8배좌우。
Motion estimation(ME) module is one of the most important module in digital Image/Video processing, its quality has directly respect to the Image/Video processing efficiency. Fast and efficient ME algorithms are desired in many applications.Using CUDA(Compute Unified Device Architecture) technology can resolve this problem very well. CUDA is a parallel computing architecture developed by NVIDIA. In this paper, we finished the full search algorithm(FS) and other ME algorithms using the CUDA technology. Experiment results show that CUDA-based ME algorithms can be more than 8 times as fast as the CPU-based ME algorithms.