The content of this article comes from the "Surveying and Mapping Bulletin" No. 6 in 2024, drawing review number: GS Jing (2024) No. 1024
Construction waste classification method based on multi-feature combination
ZHANG Dai Xinyue1, LIU Yang1,2, GAO Siyan3,4
1. School of Surveying, Mapping and Urban Spatial Information, Beijing University of Civil Engineering and Architecture, Beijing 102616, China; 2. Beijing Advanced Innovation Center for Future Urban Design, Beijing University of Civil Engineering and Architecture, Beijing 100044; 3. Zhengyuan Geographic Information Group Co., Ltd., Beijing 101300, China; 4. Beijing Engineering Research Center for Safety Evaluation and Operation Supervision of Smart Pipe Network, Beijing 101300, China
Funds: National Natural Science Foundation of China (42271478); National Key R&D Program of China(2018YFC0706003)
Key words: construction waste, multi-feature combination, random forest, Hyperspectral imagery
Citation format: Zhang Dai Xinyue, Liu Yang, Gao Siyan. Construction waste classification method based on multi-feature combination[J]. Bulletin of Surveying and Mapping, 2024(6): 59-64.doi: 10.13474/j.cnki.11-2246.2024.0611
Abstract:The rapid development of cities produces a large amount of construction waste, which leads to urban pollution and the phenomenon of "garbage siege". In this paper, a certain area in Beijing was taken as the study area to study the effect of multi-feature combination method on the accuracy of construction waste classification in hyperspectral images, and the spectral data measured by ASD QualitySpecTrek handheld spectrometer and Zhuhai-1 hyperspectral images were used to carry out construction waste classification experiments. Ninety characteristic variables, such as spectral features, vegetation index, water index and texture features, were extracted, and the importance of the features was ranked by the method of reducing the average impurity. The experimental results show that the random forest classification method using multi-feature combination has better effect than the traditional random forest method, with an overall classification accuracy of 85.86% and a Kappa coefficient of 0.80, while the original random forest method has an overall classification accuracy of only 83.48% and a Kappa coefficient of 0.75, which indicates the effectiveness of the multi-feature combination random forest algorithm.
About author:ZHANG Daixinyue (1998—), female, master's student, main research direction is geographic information system and hyperspectral remote sensing. E-mail:[email protected]
First instance: Yang Ruifang review: Song Qifan
Final Judge: Jin Jun
information