摘要： 现有的图像校正技术仅对触点坐标进行校正，存在损失大量操控面信息，限制复杂交互方式的缺点。为此，利用图形处理器(GPU)的可编程性和图形图像计算方面的优势，由GPU完成图像校正和处理过程，以此提高图像处理速度。为使系统支持更复杂的操控与交互，改进图像校正和处理流程，实现图像的整体校正，在保留操控面全部信息的同时减少CPU占用资源。理论分析与实验结果验证了该方法的有效性。

关键词: 图形处理器, 多点触控, 图像校正, 人机交互

Abstract: Current solution only calibrates coordinates the points to reduce the CPU resource occupation. It loses large amount of control surface information and limits the development of more complex interactions. As to increase the speed of image processing, by exploiting the advantages in computing area of modern programmable Graphics Processing Unit(GPU), this paper presents a solution to transport the image calibration and image processing into GPU. In order to support more complex manipulation and interaction, the process of image calibration and processing is improved. With higher process speed and less CPU resource, the whole image is calibrated to keep all the information of control surface. The analysis and experiments prove its good practicability and effectiveness.

Key words: Graphics Processing Unit(GPU), multi-touch, image calibration, human-computer interaction

中图分类号: 

• TP391.41