基于神经网络与贝叶斯信息融合的小白菜成熟度检测方法  

Detection Method of Pakchoi Maturity based on Neural Network and Bayesian Information Fusion

作　　者：梁帆[1] 孙更强 王宇 杨莉莉[1] 崔世钢[1] 吴兴利[1] 田立国[1] 

机构地区：[1]天津职业技术师范大学/天津市信息传感与智能控制重点实验室,天津300222 [2]青岛世园(集团)有限公司,山东青岛266000 [3]山东雷士光电工程技术有限公司,山东青岛266000

出　　处：《湖北农业科学》2016年第1期202-204,241,共4页Hubei Agricultural Sciences

基　　金：国家自然科学基金项目(61178048);国家社会科学基金项目(BFA110049);国家863计划项目(SS2013AA03120);天津市应用基础与前沿技术研究计划项目(14JCQNJC04300)

摘　　要：为了掌握智能植物生长柜中小白菜的成熟情况,便于对柜内环境参数实现智能控制,提出了利用小白菜的外部形态特征,特别是提取根系形态特征并将其与地上部分形态特征相结合来检测小白菜成熟度的方法。通过Matlab图像处理工具箱对采集的小白菜图像进行阈值分割和特征提取,然后将小白菜上、下两部分的形态特征数据作为训练样本,分别建立对应的神经网络成熟度检测模型,并将神经网络检测值利用贝叶斯理论来对其进行信息融合,从而进一步提高神经网络模型检测的准确性。In order to better grasp the maturity of pakchoi in the plant growth cabinet to realize the intelligent control of the environment in the cabinet, the method of detecting the pakchoi maturity was proposed by using the external morphological characteristics of pakchoi ,especially the extraction of root morphological characteristics, and combining the above ground morphological features with the root morphological characteristics. The threshold segmentation and feature extraction were conducted for the gathered pictures of pakchoi by the image processing toolbox of Matlab, and then the data of the above ground morphological features and the root morphological characteristics of pakchoi were taken as training samples to establish two neural network maturity detection modle of the ground and underground. Finally, the information fusion of the maturity by Bayesian theory was done, and which could further improve the detection accuracy of the neural network model.

关 键 词：图像处理 神经网络 成熟度 贝叶斯理论 信息融合 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程] TP274.5[自动化与计算机技术—控制科学与工程]