中文摘要：

引导滤波算法被大量用于图像处理领域中,在去雨雪、去雾、前景提取、图像去噪、图像增强、级联采样等方面有很好的处理效果。但是对于实时应用,软件实现难以满足需要。提出了在SDSoC环境下利用软硬件协同开发策略实现引导滤波硬件加速。通过在SDSoC开发环境中调试C语言代码实现引导滤波算法,并将其中影响性能的函数用Xilinx公司开发的Zedboard开发版硬件实现。在设计中,采用了流数据的方法、PS（Processing System）端和PL（Programmable Logic）端协同开发策略,以及软硬件并行、流水线优化等优化方法,提高了加速器的整体性能。实验结果表明,提出的软硬件协同的引导滤波加速器加速比可达16。

英文摘要：

The guided filter is widely used in image processing fields such as removal rain and snow, removal smog, foreground extraction, image denoising, image enhancement, cascade sampling and so on. But it is hard to satisfy real time requirement base on software. In the proposed method, the guided filter is speed up based on SDSoC environment using hardware/software co-design approach. Guided filter algorithm is implemented by debugging C language in SDSoC environment, and then some functions which impact the performance are implemented by hardware using the Zedboard from Xilinx company. The algorithm uses the method of data stream, PS（Processing System） and PL（Programmable Logic） co-design strategy, parallelism of hardware and software, pipeline optimization, reconstruction sequence of algorithm and the combination of line buffer and window buffer. The experimental results show that the accelerator＇s speed-up ratio can be 16.