Abstract: This paper presents a FPGA based fast Hough transform for line detection.The input edge image is divided into several directions by using classified filter，so every direction can be processed in parallel.And for each direction，a pipelined structure is used for line extraction.A parameter storage and search strategy is proposed，with which 2-step histogram search strategy is introduced in the purpose of reducing the storage requirement.Experiments show that the algorithm is efficient，low storage requirement and real time.

Key words: Hough transform, Field Programmable Gate Arrays（FPGA）, classified filter, pipelined structure

摘要： 在FPGA上设计并实现了一种用于直线检测的快速Hough变换方法。使用分类滤波器把直线目标分成多个方向，使多个方向上的运算在空间上实现了并行处理；在每个方向上，设计实现了一种用于Hough变换的流水线处理结构；提出了一种基于直方图统计的两阶段搜索算法。大量的实验验证了提出的Hough变换实现方法的可行性，结果证明该方法占用空间少，实时性高。

关键词: 霍夫变换, 可编程逻辑门阵列, 分类滤波器, 流水线结构