摘要
采用数码相机拍照法对狼毒盖度进行了研究。选择过绿特征区分绿色植被和非绿色植被,结合亮度特征和面积阈值从非绿色植被中自动提取出狼毒花,并提出一种基于行程标记的孔洞填充算法对狼毒花中的孔洞进行了填充。选取10张照片对提取结果进行精度评价,Kappa系数为0.80,结果发现:对狼毒茎叶和牧草之间,不宜采用过绿特征进行区分;在HLS色彩空间下,根据色调特征自动提取狼毒茎叶效果一般;对20张照片中的狼毒花的面积和茎叶面积进行回归分析,发现二者表现出了很好的线性关系;根据回归模型计算茎叶面积效果较好;为了得到更准确的狼毒茎叶面积,获取照片时最好处于狼毒的盛花期。
The coverage of stellera was estimated based on the digital images. The excess green operator was chosen to classify green and non-green vegetation. The stellera flowers were automatically extracted from non-green vegetation using brightness feature and area threshold. The holes in the flowers were filled using run length marking algorithm. Ten photos were randomly selected for evaluating the classification precision and the Kappa coefficient was 0.80. The excess green was not sensitive between stellera leaves and grass. The precision can improve when the hue feature is used to discriminate the stellera leaves under the HLS color space. The regression analysis was carried out and the linear equation could express the relationship best between the stellera flower areas and leaves areas. The regression simulation method can get higher estimation precision of the stellera leaves.
出处
《地球科学进展》
CAS
CSCD
北大核心
2009年第7期776-783,共8页
Advances in Earth Science
基金
中国科学院"西部之光"人才培养计划项目"黑河上游毒草的遥感监测与空间分布规律研究"(编号:CACXO728501001)
中国科学院西部行动计划(二期)项目"黑河流域遥感-地面观测同步试验与综合模拟平台建设"(编号:KZCX2-XB2-09-03)
国家重点基础研究发展计划项目"陆表生态环境要素被动遥感协同反演理论与方法"(编号:2007CB714401)联合资助
关键词
狼毒
过绿特征
孔洞填充
Kappa系数
Stellera chamaejasme L.
Excess green
Hole filling
Kappa coefficient.