摘要
文中使用Raspberry Pi开源硬件平台搭建了水下机器人视觉系统,通过Open CV对所要搜寻的特定目标物体进行Haar分类器训练,AUV在进行特定目标搜寻任务时加载训练文件,并通过所开发程序在嵌入式视觉系统上完成了水下特定目标的搜寻任务。文中对训练流程进行了分析,并对分类器做出了评价,通过实际的水下实验,验证了所用训练分类器具有较低的误检率和较高的检测率,同时,所搭建的视觉系统具有较好的实时性。
The realization of autonomous underwater vehicle (AUV) identification of underwater specific targets has important practical value and research significance. This paper uses the Raspberry Pi open source hardware plat- form to build underwater robot vision system, and the Haar classifier training for specific target object to be searched by OpenCV. The AUV automatically loads the file in performing a specific task and complete the specific target of un- derwater search task through the developed program in the embedded vision system. This paper analyzes the training process and evaluated the classifier. Actual underwater experiment shows that the classifier generated by training has higher detection rate and lower error rate. The good real - time performance of the proposed scheme is also verified.
出处
《电子科技》
2017年第2期130-134,共5页
Electronic Science and Technology