基于FPGA的浮点算法在图像处理中的应用

Application of a Floating-point Algorithm Based on FPGA in Image Processing

针对数字图像本身存在的特点,提出了一种基于FPGA的浮点运算方法.该方法根据数字图像中像素点的坐标值和灰度值均为正整数的特点,利用FPGA中较易实现的定点乘法、加减运算和移位操作来实现浮点运算.这种浮点运算方法能够克服传统的浮点运算结构复杂,延时长,难以保证结果的实时性等严重不足.该算法已成功应用于以XC2S200-5PQ208为核心处理器的实时图像消旋系统中,并用ModelSim SE仿真软件进行仿真.实验结果表明,该算法原理简单,速度快,精度可调,适于实时图像处理.

In view of some characteristics of digital image, a new method of floating-point operation is implemented on a ordinary FPGA chip. On the basis of the characteristic of digital image that the coordinate and the gray level of pixel are positive integer, the fixed-point multiplication, additlon/subtraction operations and shift operation which are much easier to be implemented are used to complete floating-point operation. This method of floating-point operation can get over the inadequacy of complex structure ,long delay period, hard to assure the real-time processing. This floating-point algorithm is applied successfully to a system for real-time canceling image rotation on basis of that the center processor is XC2S200 - 5PQ208, and it is emulated with ModelSim SE. The experimental results show that the method has advantages of simplicity, high speed and precision adjustable and it is suit to real-time image processing.