文献类型：期刊文献

中文题名：智能森林多状态视觉监控系统的设计与实现

英文题名：Design and Implementation of Multi-state Intelligent Visual Surveillance System of Forest

作者：李彦[1]

第一作者：李彦

机构：[1]新乡学院教务处

第一机构：新乡学院

年份：2014

卷号：22

期号：1

起止页码：136-138

中文期刊名：计算机测量与控制

外文期刊名：Computer Measurement & Control

收录：CSTPCD;;北大核心:【北大核心2011】；

语种：中文

中文关键词：遥感图像;计算机视觉;病虫害识别

外文关键词：remote sensing image; computer vision; plant diseases and insect pests identification

摘要：针对现有森林监测系统对森林的多样性事件识别功能不足的问题,设计了一套基于计算机分布式视觉模型的森林病虫害监控系统;监控系统设计了底层控制站CCD森林视觉信息采集模块和DP总线信息传递模块,准确、实时地检测出虫害、干旱、风化等森林多样性事件,增加了系统功能;利用遥感森林监控图像的均值作为初始背景;根据不同森林灾害事件获取背景图像进行实时更新;利用森林光谱特征就距离算法获取差分图像,并进行形态学处理;根据DSP处理器得到的二值图上植物目标的特征,判断发生了哪类森林灾害事件;针对该系统测试表明,系统在对高空采集的遥感森林图像的病虫害监测中,对红粉、白粉、黑星和白板的分类正确率达100%,对干旱和风化等灾害事件的识别正确率达80.00%和93.33%,适用于信息融合型的森林智能监控系统应用。
In view of the existing forest pest and disease monitoring system on disease and insect diversity event the problem of insuffi cient recognition, and designed a set of the forest pest monitoring model based on distributed computer vision system. The underlying control station monitoring system design CCD visual information collection modules and forest DP bus communication module, accurate, real time detection of pests, drought, weathering and other forest diversity, strengthen the function of system. Composed of DSP central processing module. Use more natural forest monitoring image mean value as the initial background; According to the different forest disasters access to real--time updates the background image; Distance algorithm to obtain difference images using forest spectral characteristics, and morpho logical processing; Based on DSP processors are binary plants on the drawing of the characteristics of the target and determine what type of forest disasters happened. Experimental results show that the system for the system of forest remote sensing image in the high altitude collec tion of plant diseases and insect pests monitoring, the stars red pink, white, black and white board was 100%, and the classification accuracy of downy mildew and disease--free classification accuracy of 80.00% and 80. 00%, applied to information fusion of intelligent monitoring system of forest.