多尺度3D智慧城市建筑智能检测方法创新研究

Research on the innovation of multi-scale 3D smart city building intelligent detection methods

为应对智慧城市、智能导航、自动驾驶的快速发展,提高3D智慧城市建筑检测。研究建立基于时空数据的多尺度模型,以应用于城市建筑检测。结果表明,将平均坡度阈值设为53°,以识别和去除小植被,提高建筑检测精度。当特征阈值为62时,误差最小;当特征阈值为78时,误差较特征阈值62仅增加0.1%,因此建议将特征阈值选取范围设定为62~78,以保证建筑模型的检测精度。建筑物检测的平均完整率、正确率和精度分别为88.9%、94.1%和90.6%,在区域2中的检测性能较好。

In order to cope with the rapid development of smart cities,intelligent navigation,and autonomous driving,3D smart city building detection is improved.A multi-scale model based on spatio-temporal data was established for urban building inspection.The results showed that the average slope threshold was set to 53°to identify and remove small vegetation and improve the accuracy of building detection.When the feature threshold was 62,the error was the smallest.while when the feature threshold was 78,the error only increased by 0.1%compared to the feature threshold 62.Therefore,it was recommended to set the feature threshold selection range to 62~78 to ensure the detection accuracy of the building model.The average completeness,accuracy,and accuracy of building detection were 88.9%,94.1%,and 90.6%,respectively,indicating good detection performance in region 2.