一种图像预处理结构及典型算法的FPGA实现

An Image Pre-processing Structure and the FPGA Implement of the Typical Algorithms

图像滤波和边缘检测等预处理算法是视觉导航系统中道路检测和车辆检测等复杂视觉处理的前提,其性能和处理时间直接影响了后续图像处理的性能及视觉系统的整体响应时间。本文给出了一种基于FPGA的流水线图像预处理结构,该结构具有良好的扩展性,并能克服因级联而产生的时序问题。基于这种结构,在FPGA中实现了典型的LoG边缘检测算法与中值滤波算法。实验表明了这种预处理结构的有效性和算法的可行性,并能满足视觉处理的实时性要求。

Image pre-processing such as image filtering and edge detection is an important stage in complex vision tasks,such as land mark and vehicle detection in vision navigation systems.The performance and processing time will directly affect the subsequent processing and the total response time of a vision system.This paper presents a pipelined image pre-processing structure based on FPGA.The proposed structure is scalable and can eliminate the effect of the timing problem.A typical Median Filter and a LoG edge detection algorithm are implemented in FPGA with a convolution structure based on this structure.Experiment shows that the presented structure and the typical algorithms are effective in real-time image preprocessing applications.