马尾松毛虫危害下的马尾松纹理特征分析

Texture Analysis of Pinus massoniana under Dendrolimus punctatus Damage

以福建省沙县为研究区,以地面调查数据为辅助,应用Landsat 8 OLI多光谱遥感数据提取健康、轻度、中度与重度4个虫害等级危害下的马尾松纹理特征,借助空间距离法、相关系数法、光谱角制图法分别建立虫害等级区分标准,定量描述4种虫害等级下的马尾松纹理特征,对比不同分析结果,挖掘松毛虫危害下的林分纹理特征响应机制。结果表明：（1）利用单因素方差分析法选定Mea1、Mea4、Mea7等3个纹理特征量作为具显著性差异纹理量进行纹理特征分析;（2）对比3种方法的区分精度大小：相关系数法（80.0%）〉光谱角制图法（75.0%）〉空间距离法（60.0%）,对比准确率大小：相关系数法（91.2%）=光谱角制图法（91.2%）〉空间距离法（85.0%）;（3）通过构建具显著性差异纹理量可以较有效地提取马尾松毛虫害信息;虫害等级跨度越大,基于纹理特征提取马尾松信息效果越好。

Taking Shaxian County as the study area,the texture features of healthy,lightly-infested,moder-ately-infested and heavily-infested Pinus massoniana were extracted based on the field survey data and landsat 8 OLI images.Then space distance,correlation coefficient and spectral angle mapping were used to establish the distinguishing rule and quantitatively describe the four texture features.Finally,comparing different analysis re-sults,the texture response mechanism of Dendrolimus punctatus damage was found.The results showed that：（1） by means of one way ANOVA Mea 1,Mea 4 and Mea 7 were selected as the texture measures with significant difference;（ 2） the division accuracy of these methods was compared and the result showed the order of the distingnishing accuracy of these methods：correlation coefficient（80.0%）〉spectral angle mapping（75.0%）〉spatial distance （60.0%）,and that of the contasting accuracy,correlation coefficient（91.2%）=spectral angle mapping （91.2%）〉spatial distance （85.0%）;（3） the information of Dendrolimus punctatus walker damage can be extracted more effectively by constructing a way to choose remarkable different texture measures.And the larger the span of levels between pests are,the better the performance of extracting Dendrolimus punctatus walker damage’ s information is.