基于BP神经网络解决小麦群体特征的图像理解问题  

Solving Understanding Problem of the Wheat Group Images Based on BP Neural Network

作　　者：索兴梅[1] 肖波[2] 白中英[3] 李少昆[4] 赵双宁[4] 

机构地区：[1]中央民族大学计算机科学与技术系,北京100081 [2]中央民族大学网络中心,北京100081 [3]北京邮电大学计算机与科学技术学院,北京100876 [4]中国农业科学院作物育种栽培研究所,北京100081

出　　处：《中央民族大学学报（自然科学版）》2003年第1期53-60,共8页Journal of Minzu University of China(Natural Sciences Edition)

基　　金：国家"863计划"资助项目(863 306 ZD05 01 9)

摘　　要：　小麦生长发育群体图像动态信息的识别与分析能够为小麦高产田的诊断提供定量化的诊断依据.依据诊断出的作物各生长阶段的群体结构和个体指标,通过技术措施对群体发展动态进行监测调控,使其沿着高产目标的预定方向发展.本文以小麦群体绿色面积和叶面积指标信息的获取为例,应用图像分割、图像增强技术提取小麦群体图像特征,采用BP人工神经网络(ANN)方法,建立小麦图像群体特征识别自学习系统,并将其应用于小麦图像群体特征识别中,准确率在85%以上,表明利用ANN技术对小麦图像群体特征识别是可行的.Recognition and analysis of dynamic information about wheat group images during wheat growth can be taken for the base of quantitative diagnosis for highproduction wheat croplands. On the base of the group structures and the features in the various stages of wheat growth, the group development have been monitored and regulated to the direction of high production in advance. In this paper, the method has been studied. to extract the characteristic data from the example of highproduction wheat group images by using image segmentation and image enhancement. The model of artificial neural network (ANN) has been constituted and used for recognition of highproduction wheat group images.

关 键 词：小麦高产群体图像 图像分割 图像增强 BP人工神经网络 图像识别 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]