摘要
针对现有无人机图像处理系统存在处理耗时长、处理速度低等问题,该文引入边缘计算的轻量级深度学习,开展无人机图像处理系统设计研究。设计系统硬件总体架构,并在总体架构下设计图像处理模块UART接口,完成系统硬件部分设计。利用CMOS图像传感器采集无人机图像,获取图像信号,计算图像信号的饱和度和幅度,结合边缘计算的轻量级深度学习,对图像进行渲染处理,完成系统软件部分设计。实验证明,实验组系统平均处理速度在6.25~6.85幅/min范围内,而对照A组和对照B组无法达到这一水平,验证了所设计的无人机图像处理系统的有效性。
The UAV images processing system has the problems of long processing time and low speed.Therefore,this paper introduces lightweight deep learning of edge computing to carry out the design and research of UAV image processing system.The overall hardware architecture of the system is designed and the UART interface of the image processing module is designed under the overall architecture.To achieve the design of the hardware part of the system,the CMOS image sensor is used to collect UAV images,obtain image signals,calculate the saturation and amplitude of image signals,and combine the lightweight deep learning of edge computing to render the images,thereby achieving the design of the system software.The experiment proved that the average processing speed of the experimental group system was within the range of 6.25-6.85 frames per minute,while the control groups A and B were unable to achieve this level,which verified the design effect of the research on the Unmanned Aerial Vehicle image processing system.
作者
孙飞
周逞
桑培帅
SUN Fei;ZHOU Cheng;SANG Peishuai(Anhui University,Hefei 230601,China;Anhui Jiyuan Software Co.,Ltd.,Hefei 230601,China)
出处
《电子设计工程》
2025年第22期177-181,共5页
Electronic Design Engineering
