Research on the Detection Method of Hollowing and Missing for Building Exterior Walls Based on Visible and Infrared Image Fusion
The detection of hollowing and missing of building exterior walls is crucial to ensure the public safety around aging buildings in cities.The traditional artificial in-situ detection methods are time-and la-bor-consuming with safety risks.In addition,the detection results will also be affected by subjective factors such as professional experience and working status.The method of image acquisition by UAV and detection of building exterior wall defects by artificial intelligence model has become popular.However,the current research on defect detection only focuses on visible images or infrared images of a single modality,and only detect a certain defect without considering the mutual conversion between defects.To address this issue,this research combined the visible and infrared images of the building exterior wall,considered the image information from two modalities,and compared the UNet and Res-UNet models of different depths to identi-fy the building exterior wall defects in the fused images.The experimental results showed that the Res-UNet model with a depth of 4 performed excellent on the hollowing and missing of the building exterior wall.
multimodal fusionvisible and infrared image fusionexterior wall defect detectiondeep learning
NETL
NSTL
万方数据
