基于特征位置优选整合的快速城区检测算法

遥感图像中城市区域的自动分析解译是遥感对地观测领域重要的应用方向,针对自动高效城区检测的迫切需求,提出了一种基于遥感图像的城区区域快速检测算法.首先通过智能去雾处理降低薄雾气象条件对检测的干扰,然后通过快速的关键点搜寻进行城区特征位置初选,并以全局与局部约束相结合的策略筛选出高置信度的城区特征位置,最后通过高斯渲染加权的方法整合城区特征位置,并在获得的城区高斯加权图上自适应分割出最终的城区.本算法使用Google提供的卫星图像进行算法验证测试,可得到准确的检测结果.本检测算法可满足遥感图像城区检测自动化、实时化的需求,大大减小了人工判图的工作量,能够广泛应用于机载或星载平台.

A Real-Time Urban Area Detection Algorithm Based on Feature Location Optimization and Integration

According to the needs of automatic and efficient detection for the urban areas,an urban area detection algorithm is proposed in this paper.First,the intelligent haze removal processing is used to reduce the interference in detection.Second,the primary feature locations of urban are extracted by the feature points.Then with the combination of the global and local constraints,the highly reliable urban locations are selected.Finally,the urban characteristic locations are integrated by the method of Gaussian rendering weighted and the final urban areas are obtained through adaptive segmentation.The algorithm is tested using Google satellite images to get accurate results.It can meet the needs for automatic and real-time detection in the remote sensing image of urban areas and greatly reduce the workload of manual interpretation.