Abstract:

Tree species classification is an important application area in forestry remote sensing.It has widely used application scenarios in the areas of sustainable forestry management,biodiversity research,and ecological environmental protection.Based on the research results in this field after 2000 and the application practice in the forestry production process in recent years,the application of multi-source data in tree classification is summarized.From the perspectives of workflow and corresponding algorithms,the issue is analyzed and compared based on image classification and mathematical statistics.Facing the problems and challenges encountered in the application of forestry remote sensing in the classification of tree species,different working ideas based on semantic segmentation and instance segmentation were proposed,and the future development of multi-source remote sensing data fusion acquisition and hardware processing equipment was prospected.

Key words: forestry remote sensing, tree species classification, lidar data, hyperspectral data, theory application