基于FPGA的数值计算在实时图像处理中的应用

实时图像分割对于实时图像处理系统具有非常重要的意义。对于具有二次曲线边界的图像,为了确定其边界以进行实时分割,必然涉及到硬件电路上数据的开方等复杂计算,这对于一般情况下数据传输速率达到80MHz左右的实时图像处理系统来说无疑是个很大的挑战。为了完成图像分割,文章通过一种逐位比较的算法以确定图像边界,该算法的数学本质是利用二分法进行数值计算。而在具体实现方面,设计了基于FPGA的硬件电路,其中处理器内部结构采用数据通路和数据控制通路分离的方式,并通过VerilogHDL语言编程下载。最终实现了对具有二次曲线边界的图像的实时物理分割。

Application of Numerical Calculation Based on FPGA in Real-time Image Processing

Real-time image segmentation has great significance of real-time image processing. To capture the boundary of an image with conic edges, extracting roots on circuit is affirmatively involved. For real-time image processor, which designed to handle pixel data transmitting in approximately 80 MHz, it is definitely a challenge. The bit after bit comparison arithmetic with the hypostasis of bisection method is applied to real-time image processing. The hardware, a circuit including FPGA, is designed primarily. And the processor's structure inside the FPGA is programmed in advance of separate structure of data flow and controller. The purpose of real-time physics image segmentation is finally completed.