一种基于边缘计算的野外目标探测器设计与实现

Design and implementation of a wild object detector based on edge computing

为了满足监测点和监测数据快速增长的需求,在图像处理、存储和通信等技术创新的驱动下,边缘计算在野外探测应用中正在快速发展。探测器采用一款SOC的图像处理器(JL3365A)和微控制器(TM57PE20A)进行搭建,外围配置SD接口、WIFI模块、红外探测和温度传感器等功能模块。软件的控制内核充分利用微控制器的双时钟进行设计,四种工作模式的设计以最大程度的节约能耗,满足野外独立电源的需要。探测器采用红外探测和检测算法相结合的方法,背景差法实现系统唤醒时的目标检测,提高检测的准确性,帧间差分法实现运动目标的追踪检测。文件系统是基于嵌入式FatFs文件系统进行移植开发,以提高开发的效率。通过搭建测试网络进行测试,探测器实现了对野外运动目标的准确检测,并在计算边缘侧实现了数据的共享,具有较高的应用价值。

To meet the requirement for the increasing of monitoring sites and data,edge computing is developing rapidly in the wild detection with new technologies of image processing,storage and communication.The detector is designed with a SOC image processor(JL3365A)and microcontroller(TM57PE20A),including the peripheral modules such as SD interface,WIFI module,PIR and temperature sensor.The software control core takes full advantage of the dual clocks of the microcontroller to reduce energy consumption in the wild.The detector adopts the method of infrared detection and motion detection algorithm.The background difference method improves the accuracy of detection after the system wakes up.The frame difference method helps to track the motion objects.The file system is developed based on the embedded FatFs file system which can improve the efficiency of development.Through the test network,the detector can achieve great performance for wild motion detection and share the data with edge computing,which has high application value.