农业工程2025，Vol.15Issue(10)：36-41,6.DOI:10.19998/j.cnki.2095-1795.202510007
基于成像切片连续性的异质森林冠层病虫害光谱识别技术
Spectral recognition technology for heterogeneous forest canopies with diseases and pests based on imaging slice continuity
任冬焕 1杨旭照 2吉洪坤 1王园龙 1景向方1
作者信息
- 1. 濮阳市农林科学院,河南 濮阳 457000
- 2. 濮阳市林业局,河南 濮阳 457000
- 折叠
摘要
Abstract
Heterogeneous forests often contain multiple tree species,with complex hierarchical structure in their canopy layers.Differ-ent vegetation level exhibit varying reflections,scatterings,and absorptions of light,resulting in complex spectral features.Feature wavelengths selected based on a single imaging slice affect disease and pest identification accuracy.Therefore,a spectral identification technology for heterogeneous forest canopies diseases and pests based on imaging slice continuity was proposed.Continuous projection of imaging slices for heterogeneous forest canopies diseases and pests was performed.Based on absolute values of projection vectors across continuous imaging slices set,wavelengths were iteratively updated to select feature wavelengths covering continuous imaging slices.Using top N feature wavelengths with the largest projection vector,extended ratio spectral indexes were calculated and mapped to Lambe angle field of heterogeneous forest canopies disease and pest images.Based on mapping results,GAFS was used to generate spectral distribution information of heterogeneous forest canopies diseases and pests through inverse inference,completing spectral recog-nition.Test results indicated that design method performed well in recognition accuracy,and,with Kappa value remaining stable above 0.6 after 15 iterations,indicating recognition stability.关键词
成像切片/异质森林/光谱识别/连续投影算法/特征波长/扩展比值光谱指数/GAFSKey words
imaging slice/heterogeneous forest/spectral recognition/successive projections algorithm/characteristic wavelength/extended ratio spectral index/GAFS分类
农业科技引用本文复制引用
任冬焕,杨旭照,吉洪坤,王园龙,景向方..基于成像切片连续性的异质森林冠层病虫害光谱识别技术[J].农业工程,2025,15(10):36-41,6.
