Extraction Method of Urban Forest Land Information Based on Spectral Information

doi:10.13466/j.cnki.lyzygl.2021.03.015

Abstract

Abstract:

In order to study the extraction method of forest land information with more regional characteristics,some regions of Guiyang were selected as the research object,and the terrain information was preprocessed with the help of spectral information.The results showed that this method can effectively reduce the pressure of the segmentation model,and the IoU of this method was 0.943 6,which was 6.79% and 15.74% higher than U-Net method and NDVI threshold method respectively,,showing that the method is effective for extracting forest land information in urban area.

Key words: spectral information, regional characteristics, woodland extraction, normalized vegetation index, deep learning

CLC Number:

S771.8

LIN Shuangshuang, ZHONG Jiusheng, HE Xin, JIANG Li, DUAN Jiwei, DAI Renli, HE Zhiyuan. Extraction Method of Urban Forest Land Information Based on Spectral Information[J]. FOREST RESOURCES WANAGEMENT, 2021, (3): 96-100.

Figures/Tables 5

Fig.1

Fig.2

Fig.3

Tab.1

Tab.2

References 18

[1]	凌春丽, 朱兰艳, 吴俐民. WorldView-2影像林地信息提取的研究与实现[J]. 测绘科学, 2010, 35(5):205-207.
[2]	张成才, 娄洋, 等. 基于像元二分模型的伏牛山地区植被覆盖度变化[J]. 水土保持研究, 2020, 27(3):301-307.
[3]	Hasituya, CHEN Zhongxin, WANG Limin, et al. Selecting appropriate spatial scale for mapping plastic-mulched farmland with satellite remote sensing imagery[J]. Remote Sensing, 2017, 9:265. doi: 10.3390/rs9030265
[4]	程娟, 肖青, 闻建光, 等. 地物波谱数据库应用方法及遥感应用现状[J]. 遥感技术与应用, 2020, 35(2):267-286.
[5]	Oquab M, Bottou L, Laptev I, et al. Learning and transferring Mid-Level image representations using convolutional neural networks [C]// Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Columbus,OH:IEEE, 2014.
[6]	颜伟, 周雯, 易利龙, 等. 森林类型遥感分类及变化监测研究进展[J]. 遥感技术与应用, 2019, 34(3):445-454.
[7]	李青, 焦李成, 周伟达, 等. 基于向量投影的支撑向量预选取[J]. 计算机学报, 2005, 28(2):145-152.
[8]	刘巍, 吴志峰, 骆剑承, 等. 深度学习支持下的丘陵山区耕地高分辨率遥感信息分区分层提取方法[J]. 测绘学报, 2021, 50(1):105-116.
[9]	邹橙, 杨学志, 董张玉, 等. 基于GF-2遥感影像的一种快速水体信息提取方法[J]. 图学学报, 2019, 40(1):99-104.
[10]	刘筱笛. 贵阳市周边地区特色村落规划方法研究[D]. 西安:长安大学, 2012.
[11]	周晓芳. 贵阳市发展都市农业旅游探讨[J]. 山地农业生物学报, 2003, 22(1):66-69.
[12]	白洁, 葛全胜, 戴君虎. 贵阳木本植物物候对气候变化的响应[J]. 地理研究, 2009, 28(6):1606-1614.
[13]	肖红伟, 肖化云, 王燕丽. 贵阳大气降水化学特征及来源分析[J]. 中国环境科学, 2010, 30(12):1590-1596.
[14]	冯小铭, 郭坤一, 王敬东. 对中国城市环境地质工作的思考[J]. 安全与环境工程, 2001, 8(4):1-4.
[15]	Tucker C J. Red and photographic infrared linear combinations for monitoringvegetation[J]. Remote Sensing of Environment, 1979, 8(2):127-150. doi: 10.1016/0034-4257(79)90013-0
[16]	蒲智, 刘萍, 杨辽, 等. 面向对象技术在城市绿地信息提取中的应用[J]. 福建林业科技, 2006, 33(1):40-44.
[17]	Ronneberger O, Fischer P, Brox T. U-Net:Convolutional networks for biomedical image segmentation [C]// Proceedings of International Conference on Medical Image Computing and Computer-Assisted Intervention.Munich:Springer, 2015:234-241.
[18]	赵松乔, 陈传康, 牛文元. 近三十年来我国综合自然地理学的进展[J]. 地理学报, 1979, 34(3):187-198. doi: 10.11821/xb197903001

类型	公式	公式含义
IoU	(A∩B)/(A∪B) (2)	A为所提取林地图斑面积; B为对应的实际林地图斑面积
Kappa	(Po-Pe)/(1-Pe)(3)	Po表示总体分类精度; 假设每一类的真实样本个数分别为a₁,a₂,…,a_m 而预测出来的每一类的样本个数分别为b₁,b₂,…,b_m 总样本个数为n 则有:Pe=(a₁×b₁+a₂×b₂+…+a_m×b_m)/(n×n)
OA	TP/(TP+FP) (4)	TP表示正确分类象元数; FP表示错误分类象元数

提取方法	OA/(%)	Kappa	IoU
本文方法	98.19	0.9645	0.9436
U-Net方法	96.71	0.9114	0.8757
NDVI阈值方法	93.10	0.8302	0.7862