新农村环境垃圾监测信息处理系统研究

发布时间：2017-12-27 05:00

本文关键词：新农村环境垃圾监测信息处理系统研究　出处：《南京理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：为了对新农村环境垃圾问题进行有效监测,本文以新农村环境垃圾监测信息处理系统为研究对象,设计了系统的软硬件架构,开发了一种基于HOG特征的SVM机器学习目标检测算法,实现了对新农村街道上的垃圾进行检测。首先针对新农村街道卫生的特点,分析了系统的功能要求和技术指标,设计了新农村环境垃圾监测信息处理系统的总体架构。开发了能保证垃圾监测的需要和以G-T40N双核CPU为核心的硬件系统,完成了基于可见光相机的图像采集模块电路,设计了以SLM630无线通讯模块为核心的数据传输电路,预留了硬件系统的可扩展功能接口。分析了 HOG特征从正负样本垃圾图像提取的过程,论述了 SVM机器学习算法能进行分类的原理,研究了提取的HOG特征与SVM机器学习算法之间的理论数学关系,设计了目标检测算法的软件流程;根据信息采集系统的数据需要,设计了数据采集和相机变倍的软件流程,开发了无线数据传输软件模块,满足了后端能实时查看相机的拍摄画面信息。最后对所开发的环境垃圾监测信息处理系统进行了调试和实验分析,结果表明系统能满足对新农村垃圾问题的监测需要,提出的机器学习目标检测算法,通过实验验证了算法的可行性,为新农村环境垃圾监测问题提供了一种参考方法。分析了算法训练出的SVM目标分类器ROC曲线图,得到分类器的准确率为95.56%,提出了可以继续提高性能的方法。
[Abstract]:In order to effectively monitor the environment of the new rural garbage problem, based on the new rural garbage environmental monitoring information processing system as the research object, design the hardware and software architecture of the system, the development of a learning machine SVM target detection algorithm based on HOG features, to achieve the detection of the new rural garbage on the street. First, according to the characteristics of new rural street sanitation, the functional requirements and technical indicators of the system are analyzed, and the overall framework of the new rural environmental garbage monitoring information processing system is designed. Developed to ensure waste monitoring and G-T40N dual core CPU as the core of the hardware system, the circuit of image acquisition module visible camera based on the data transmission circuit based on SLM630 wireless communication module as the core of the design, set aside a hardware system extension function interface. Analysis of the HOG features from the process of positive and negative samples of spam image extraction, discusses the SVM machine learning algorithm to classify the principle, on the extraction of HOG features and SVM machine learning theory the mathematical relationship between algorithms, design the target detection algorithm of software process; according to the information collection system of data, design data collection and the camera zoom software development process, software wireless data transmission module, to meet the end can view real-time camera image information. At the end of the environmental waste monitoring information processing system developed for debugging and experimental analysis, the results show that the system can meet the needs of the new rural garbage problem monitoring, machine learning target detection algorithm is proposed, through the experiment to verify the feasibility of the algorithm, provides a reference method for the new rural environmental monitoring garbage problem. The ROC curve of the SVM target classifier trained by the algorithm is analyzed. The accuracy rate of the classifier is 95.56%, and the method that can continue to improve the performance is proposed.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：X84;TP274

【相似文献】