| 基于GPU的弹性图像配准方法 |
| |
| 引用本文： | 张加万,杨甲东,孙济洲,张红颖.基于GPU的弹性图像配准方法[J].天津大学学报(自然科学与工程技术版),2008,41(8):946-950. |
| |
| 作者姓名： | 张加万 杨甲东 孙济洲 张红颖 |
| |
| 作者单位： | [1]天津大学计算机科学与技术学院,天津300072 [2]中国民航大学航空自动化学院,天津300300 |
| |
| 基金项目： | 国家自然科学基金，天津市科技攻关项目 |
| |
| 摘 要： | 通常的弹性配准技术因其计算强度大，消耗时间长，难以满足实时应用的要求．新一代图形处理器（GPU）以其用户友好的可编程性和出色的并行计算能力，为解决该问题提供了新的途径．根据GPU的自身特点，以薄板样奈插值作为变换模型，构建了弹性配准计算平台．对二维单模态和多模态的两组图像进行实验，结果表明，相比于CPU，利用GPU可以更为迅速地获得变换参数，对于大尺寸、高分辨率或者多局部形变的图像，GPU的处理速度超出CPU 1个数量级以上．
|
| 关 键 词： | 弹性图像配准 图形处理器(GPU) 薄板样条(TPS) |
GPU-Based Elastic Registration |
| |
| ZHANG Jia-wan,YANG Jia-dong,SUN Ji-zhou,ZHANG Hong-ying.GPU-Based Elastic Registration[J].Journal of Tianjin University(Science and Technology),2008,41(8):946-950. |
| |
| Authors: | ZHANG Jia-wan YANG Jia-dong SUN Ji-zhou ZHANG Hong-ying |
| |
| Institution: | ZHANG Jia-wan, YANG Jia-dong, SUN Ji-zhou, ZHANG Hong-ying ( 1. School of Computer Science and Technology, Tianjin University, Tianjin 300072, China; 2. College of Aeronautical Automation, Civil Aviation University of China, Tianjin 300300, China) |
| |
| Abstract: | It is difficult to employ the elastic registration technique in real time application due to its computation cost. With user-friendly programmability and parallel computation,contemporary graphics processing units （GPU） provide a new strategy to solve the problem above. Based on the characteristics of GPU, a platform was proposed to deal with elastic registration using thin-plate spline as the transformation model. Experiments for 2D registration of monomodal and multimodal images were performed. Results show that by using GPU transformation can be obtained more rapidly than by CPU. Particularly, the processing speed of GPU is higher than that of CPU by over one order of magnitude when dealing with images of large size and high resolution. |
| |
| Keywords: | elastic registration graphics processing units （GPU） thin-plate spline （TPS |
| 本文献已被 维普 万方数据 等数据库收录！ |
|
