Vega视景仿真中场景实时图像处理方法

Real-time Image Processing for Scenes in Vega Simulation Application

在Vega视景仿真的许多应用场合,为渲染出来的场景实时添加各种图像效果(如模糊、锐化、边缘检测等)成为一种需要.针对Vega没有提供相关图像处理模块,且通过将场景保存下来用传统方法进行处理无法保证视景仿真实时性的问题,研究了Vega的程序结构、通道回调机制,给出了对Vega场景进行实时图像处理的途径;通过OpenGL图像处理子集卷积功能和着色器这两种方法,在对场景的纹理映射过程中完成了对场景的图像处理.实际仿真表明,着色器方法充分利用了现代GPU的强大图形能力,在对2560x1024大小的场景进行锐化时,整个场景的绘制时间在3.24毫秒左右,完全满足实时性要求并具有很大的灵活性.

In Vega scene simulation applications,adding real-time graphics effects（blur,sharpen,edge detection and so on） to the scenes is necessary.In order to solve the problem that Vega does not provide correlative image processing module and tradition methods can not satisfy the real-time requirement,the program architecture and channel callback function of Vega are researched,the way of real-time image processing for Vega scene is obtained.Two methods,OpenGL Imaging Subset convolution function and Shader,are used to perform the image processing when applying scene texture mapping.The simulation experiment results show that the OpenGL Shader method takes full use of the graphic ability of modern GPU;when scene with the size 2560x1024 is sharpened,the draw time is about 3 millisecond per frame;it satisfies the real-time requirement completely and is flexible.