基于改进马尔可夫链算法的农业害虫定位研究  

Study on the location detection of agricultural insect pests based on improved Markov chain algorithm

作　　者：田源[1] Tian Yuan(Henan University of Animal Husbandry and Economy, Zhengzhou Henan 450044, Chin)

机构地区：[1]河南牧业经济学院,河南郑州450044

出　　处：《中国植保导刊》2017年第7期30-34,共5页China Plant Protection

基　　金：河南省社科联研究项目(SKL-2013-506);郑州市社科联研究项目(JX20140577)

摘　　要：为了提高农业害虫定位的效果,提出改进马尔可夫链算法。首先用非齐次马尔可夫链算法提取像素随机序列水平、垂直方向状态数等害虫特征,设定检测概率阈值,去除伪信息;接着用隐性非齐次马尔可夫链算法将害虫定位提取得到的信息以序列输入,当最大后验概率大于固定阈值的状态序列即为定位目标;最后给出了算法流程。实验仿真显示该算法定位叶片表面的农业害虫时,在准确性、最大偏移量方面较优。In order to improve efficiency of the location detection of agricultural insect pest, Markov chain algorithm was proposed to improve. Firstly, based on non-homogeneous Markov chain algorithm, target characters of insect pests,including pixel random sequence, horizontal state and vertical direction number, were extracted, and detection probability threshold was set and false information was removed. Secondly, based on hidden Markov chain algorithm, information of insect pest location extraction was input as sequence, and target state sequence was detected when the maximum of a posteriori probability was greater than the fixed threshold. Finally, the algorithm was given. Experimental simulation showed that improved Markov chain algorithm was better in aspect of accuracy and most offset as detecting agricultural insect pests on leaf.

关 键 词：非齐次马尔可夫链 隐性 特征提取 农业害虫 定位检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]