基于像素着色器的图像处理

Image Processing Based on Pixel Shader

在大量的模式识别、图像识别的系统中,会对原始采集的图像数据进行各种各样的处理,这些处理一般都是将原始图像通过二维的线性时不变系统,即作二维的卷积处理。用普通的计算机系统计算卷积处理中的加和乘效率比较低。本文介绍利用通用显示卡里的像素着色器快速完成图像处理的方法,并给出仿真结果。

In many pattern reeognition and image reeognition systems, the raw image data are processed by many ways. In these methods, the image is faith.red with two dimension linear time-invariant system by doing two dimension convolution. The efficiency of computing convolution is very low by using common computer systems. This paper discusses how to do image processing efficiently by using pixel shader. Finally, the simulation result is given.