Back to Search
Start Over
Quantitative characterization and evaluation of highway greening landscape spatial quality based on deep learning.
- Source :
- Environmental Impact Assessment Review; Jul2024, Vol. 107, pN.PAG-N.PAG, 1p
- Publication Year :
- 2024
-
Abstract
- While it is widely known that the greening landscape spatial quality plays a vital role in promoting the level of highway landscape design, there is still no consensus on how to measure it quantitatively on a large scale. Recent advancements in computer vision tasks under deep learning, along with esmerging datasets such as street view pictures, have shown the potential to overcome previous limitations, leading to a paradigm shift in research. This paper proposes a novel digital measurement technique that automatically identifies elements and greening landscape spatial features from a partial panoramic and spherical view field, considering the driver and passengers' environmental perspectives. To achieve this, we have developed a spatial quality parameter index system consisting of three dimensions and nine indicators, with two regions dedicated to plains and mountainous regions. We have also created a visualization atlas that illustrates the landscape of highway greening in both vertical and horizontal dimensions. We applied this method to the Beijing section of the Beijing-Tibet highway to guide targeted improvement strategies from different perspectives. Our findings provide strategies for enhancing scientific performance and tools for quantitative analysis in digitizing, engineering, and refining highway landscape design. • Propose a high-precision training model Panoptic-DeepLab. • Propose a digital measurement indicator system of highway greening landscape spatial quality. • Gives an atlas representation method of highway greening landscape spatial quality. [ABSTRACT FROM AUTHOR]
Details
- Language :
- English
- ISSN :
- 01959255
- Volume :
- 107
- Database :
- Supplemental Index
- Journal :
- Environmental Impact Assessment Review
- Publication Type :
- Academic Journal
- Accession number :
- 177871620
- Full Text :
- https://doi.org/10.1016/j.eiar.2024.107559