江西科学2026，Vol.44Issue(2)：319-327,9.DOI:10.13990/j.issn1001-3679.2026.02.017
一种面向机场净空风险区的识别方法
A Method for Identifying Airport Clearance Risk Zones
摘要
Abstract
Airport clearance zones are critical areas for ensuring flight safety;however,tradi-tional monitoring approaches are often inefficient,costly and lack spatial specificity.To ad-dress these limitations,an improved DeepLabV3+model integrated with a Multi-Scale At-tention Aggregation(MSAA)module is proposed to automatically extract potential risk sources,such as buildings and vegetation,from high-resolution remote sensing imagery.These extracted objects are then spatially overlaid with the statutory Obstacle Limitation Surfaces(OLS)to delineate excessive-height risk zones.Experimental results indicate that the improved model achieves a mean Intersection over Union(mIoU)of 84.97%on the test set,representing an improvement of 5.94%over the original DeepLabV3+model.In a case study of an airport in southern China,94.6%of known penetrating obstacles are successful-ly identified within the delineated risk attention zones.Moreover,the identified zone ac-counts for only 2.35%of the total clearance protection area,corresponding to a 97.6%re-duction in non-essential monitoring scope.The proposed method provides an efficient and cost-effective intelligent decision-support tool for airport clearance safety management,promoting a transition from comprehensive inspection to refined and targeted supervision.关键词
机场净空区/机场风险关注区/深度学习/DeepLabV3+/多尺度注意力/障碍物限制面Key words
airport clearance zone/airport risk attention zone/deep learning/DeepLabV3+/Multi-Scale attention/obstacle limitation surfaces分类
信息技术与安全科学引用本文复制引用
赵景强,刘为元,魏厚雄,易文浩..一种面向机场净空风险区的识别方法[J].江西科学,2026,44(2):319-327,9.基金项目
基金信息:江西省自然科学基金重点项目(20232ACB203025). （20232ACB203025）
