基于FPGA的智能服务机器人图像预处理模块设计

摘要

为了实现智能服务机器人视觉构件的可重用与可置换,减少研发中大量低水平重复工作,根据机器人模块化体系结构的设计思想,构建了一种基于FPGA的智能服务机器人图像预处理模块,根据硬件功能划分,设计了电源单元、图像采集单元、CAN通讯等单元.程序上采用了时下较为灵活的Verilog HDL语言进行设计,实现了图像采集模块、内存读写控制模块、CAN通讯模块等可移植的程序;在此平台上实现了改进的中值滤波算法和基于深度流水线的模板卷积算法.实验对比了不同滤波算法的效果,提供了选择图像预处理算法的依据.%In order to achieve the reusability and replacement ofvision-components for intelligent service robot and reduce much low-level duplicated researching work,image pre-processing module for a FPGA-based intelligent service robot is constructed according to the designing idea of modularity architecture of robot. The powder unit,image acquisition unit, CAN communication unit etc are designed according to hardware's function. While the program is designed with flexible Verilog HDL language with which transplantable image acquisition module ,memory control module ,CAN communication module etc are achieved.Revised median filter algorithm and template convolution algorithm basing on deep pipeline is implemented.By comparing different filtering algorithms'effect,the experiment provides the evidence for pre-processing algorithm of the intelligent service robot.