基于高光谱成像的不同花期薇甘菊特征分析及监测模型研究

Feature analysis and model monitoring of different florescences of Mikania micrantha based on hyper-spectrum imaging

薇甘菊是世界上危害巨大的杂草之一,入侵能力强,严重威胁当地生态系统.使用SOC710VP成像光谱仪获取薇甘菊不同花期高光谱数据,选取与薇甘菊不同花期显著相关的参数,建立不同花期监测模型.结果表明,盛花期和未开花期光谱反射率存在明显“绿峰”和“红谷”,整体值大小依次为盛花期、未开花期、枯花期;一阶微分曲线在绿光波段、红边,未开花期、盛花期和枯花期均存在明显波峰,且值的大小依次为未开花期、盛花期、枯花期;在750~900 nm,盛花期和未开花期波峰、波谷出现波段范围大致相似,枯花期出现3个波峰,峰值依次减小;薇甘菊不同花期与光谱反射率及光谱一阶微分在529~671、734~744、770~786、791~796 nm和833~838 nm极显著相关,以NDVI、RVI、GI、DVI和HI特征参数构建的多元线性回归模型的拟合效果最佳.根据薇甘菊高光谱特征参数建立的监测模型,可有效地监测薇甘菊不同花期,研究结果可用于薇甘菊发生发展的监测.

Mikania micrantha is one of the most harmful weeds in the world,which has strong invasion ability and seriously threatens the local ecosystem.Different flowering monitoring models were established,by using SOC710VP portable imaging hyperspectral spectrometer to obtain hyperspectral data of different flowering stages of Mikania micrantha,and selecting parameters significantly related to different flowering stages of Mikania micrantha.The results show that there was obviously a“green peak”and a“red valley”in the spectral reflectance respectively at the blooming and non-blooming stages,while the“green peak”and“red valley”disappeared gradually at the withering stage,and the overall values were in the order of flowering,non-blooming and withering stages;the first-order subdivision curve was in the green light band and red edge while there were obvious“wave peaks”in the three stages of non-blooming,blooming and withering with their values in the order of non-blooming,blooming and withering stages.At 750—900 nm,the wave peaks and wave troughs in blooming and non-blooming periods were similar in frequency ranges,and there were three wave peaks in the withering period with their values decreasing gradually.In different flowering periods,the spectral reflectance and the first-order differential spectrum were significantly correlated at 529—671 nm,734—744 nm,770—786 nm,791—796 nm and 833—838 nm;and the fitting effect of the multiple linear regression model constructed with NDVI,RVI,GI,DVI and HI was the best.The monitoring model,constructed with the hyperspectral characteristic parameters of Mikania micrantha,can effectively monitor different flowering stages of Mikania micrantha,and the research results can be used to monitor the occurrence and development of Mikania micrantha.