多变分模态分解下的湿地植被高光谱识别特征波长优选与模型研究  

Modelling Wetland Vegetation Identification at Multiple Variational Mode Decomposition

作　　者：李璇 袁希平 甘淑[2,3] 杨敏 龚伟圳 彭翔 LI Xuan;YUAN Xi-ping;GAN Shu;YANG Min;GONG Wei-zhen;PENG Xiang(West Yunnan University of Applied Sciences,Key Laboratory of Mountain Real Scene Point Cloud Data Processing and Application for Universities,Dali 671006,China;Faculty of Land Resources and Engineering,Kunming University of Science and Technology,Kunming 650093,China;Application Engineering Research Center of Spatial Information Surveying and Mapping Technology in Plateau and Mountain-ous Areas Set by Universities in Yunnan Province,Kunming 650093,China)

机构地区：[1]滇西应用技术大学/云南省高校山地实景点云数据处理及应用重点实验室,云南大理671006 [2]昆明理工大学国土资源工程学院,云南昆明650093 [3]云南省高校高原山区信息测绘技术应用工程研究中心,云南昆明650093

出　　处：《光谱学与光谱分析》2025年第3期601-607,共7页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(62266026);云南省科技厅基础研究专项(202201AU070108)资助。

摘　　要：高光谱数据以其高维度为特征,拥有更丰富的地物信息。在植被分类中,这种高维度数据为提升分类准确性和精度提供了更多机会。传统的特征波长建模往往因输入变量过多而导致分类精度不佳。为了克服这一问题并提高模型对湿地植被细微光谱差异的捕捉能力,以洱海东岸海滨作为研究区域展开探索,测取了3种典型湿地植被(菰、芦、槐叶蘋)的高光谱数据作为目标样本。对样本光谱曲线进行SG平滑后作为原始光谱(OS)、对原始光谱进行包络线去除变换(CR)、一阶微分(FD)并分析其光谱特征;再将原始光谱通过变分模态分解(VMD)为8个尺度。接着,用竞争性自适应重加权(CARS)算法选择出的波长作为特征波长。最后,利用寻找出的最佳参数组合放入经贝叶斯算法优化的支持向量机(Bayes-SVM)进行建模。结果表明:CARS算法提取的特征波长数量减少,且大都分布于植被的吸收特征区间内,降维效果显著;经过分解后的第4模态构建的模型(S_(4)-CARS-Bayes-SVM)分类效果最好,其精确率PR为0.9333,召回率RR为0.8889、F1分数为0.8963、AUC值为0.9286,即此模型具有很强的鲁棒性以及识别性能。Hyperspectral data are characterized by high dimensionality and richer feature information.This high-dimensional data provides more opportunities to improve classification accuracy and precision in vegetation classification.Traditional feature wavelength modelling often results in poor classification accuracy due to too many input variables.To overcome this problem and improve the ability of the model to capture the subtle spectral differences of wetland vegetation,this paper explores the east coast of the Erhai Lake as the study area and hyperspectral data of three typical wetland vegetation(Mizuno,Ruscus,and Sophora japonica)are measured as the target samples.The sample spectral curves were SG smoothing as original spectra(OS),continuum removal transform(CR),and first-order differentiation(FD)and analyzed for spectral features;then,the original spectra were decomposed by Variational mode decomposition(VMD)into 8 scales.Next,the wavelengths selected by the Competitive adaptive reweighted sampling(CARS)algorithm were used as the characteristic wavelengths.Finally,the best combination of parameters found was used to put into the Bayesian algorithm optimized support vector machine(Bayes-SVM)for modeling.The results show that the number of feature wavelengths extracted by the CARS algorithm is reduced,and most of them are distributed in the absorption feature intervals of vegetation,and the effect of dimensionality reduction is significant;the model constructed by the 4th mode after decomposition(S_(4)-CARS-Bayes-SVM)has the best classification effect,with a precision rate(PR)of 0.9333,a recall rate(RR)of 0.8889,an F1 score of 0.8963,and AUC value of 0.9286,i.e.,this model has strong robustness as well as recognition performance.

关 键 词：光谱学 湿地植被 变分模态分解 特征波长 支持向量机 贝叶斯算法 

分 类 号：O433[机械工程—光学工程]