An Efficient Implementation of the One-Dimensional Hough Transform Algorithm for Circle Detection on the FPGA
An Efficient Implementation of the One-Dimensional Hough Transform Algorithm for Circle Detection on the FPGA
复制标题
DOI：
10.1109/candar.2014.32
复制
发表时间：
2014-12
期刊：
影响因子：
--
通讯作者：
Xin Zhou;Yasuaki Ito;K. Nakano
中科院分区：
文献类型：
--
作者：
Xin Zhou;Yasuaki Ito;K. Nakano
本文的主要贡献是提出了一种仅使用一维参数空间的Hough变换算法在Xilinx Virtex-7上的高效圆检测实现。我们使用398片DSP48E1和309块18Kbit的RAM实现了该电路。实验结果表明，该结构的工作频率为181.812 MHz。对于400×400大小的边缘图像，我们的电路最多可以在970434个时钟周期内执行，即5337.568μS，与在中央处理器上顺序实现相比，我们的实现获得了大约189倍的加速倍数。
The main contribution of this paper is to present an efficient implementation of the Hough transform algorithm that uses only one-dimensional parameter spaces for circles detection on a Xilinx Virtex-7 FPGA. We implemented the circuit using 398 DSP48E1 slices and 309 block RAMs with 18Kbits. The experimental results show that the architecture runs in 181.812MHz. For an edge image of size 400 × 400, our circuit can perform in at most 970434 clock cycles, i.e., 5337.568μs. Our implementation attains a speed-up factor of approximately 189 over the sequential implementation on the CPU.