Research On Building Change Detection Methods Based On UAV Images

With the continuous improvement of the economic level,the speed of urban expansion in China has also become faster and faster.At the same time as the expansion of the city,there have been many illegal buildings.These illegal buildings do not meet the requirements of building regulations,so they are very dangerous.If manual inspection is used to carry out the inspection of illegal buildings,a large amount of manpower is required.UAV is a rapidly developing technology in recent years.UVA can acquire high-resolution images of experimental areas in a short time,so they are often used in conduct power inspections,disaster investigations and other fields.UVA can obtain images conveniently and quickly,and the change detection technology can be used to detect building changes.However,how to use the high-resolution UVA image to detect new illegal buildings is a challenge in the field of change detection.Aiming at the above problems,this paper mainly studies the building change detection method using UAV images.The specific research content is as follows.(1)In-depth research and summary of UAV photogrammetry principle technology and deep learning theory are achieved;(2)The existing object-oriented active contour change detection technology is researched,and extended it to UAV ultra-high-resolution images building change detection.Firstly,the UAV image was segmented by statistical region merging,and the active contour model was used to detect the change.Experimental results show that this method can be used for drone building change detection,but it is greatly affected by segmentation scale and lighting changes.(3)The convolutional Siamese deep learning network is studied and extended to drone image buildings scene change detection.By constructing changed and no-changed samples,the deep network is trained to detect building scene change.Experimental results show that the network can achieve high-precision detection of building scene changes,and is less affected by factors such as illumination and registration errors.Through the above research,this article has realized an automatic building detection method based on drone images.The study can greatly reduce the workload of field work and manual investigation,and at the same time provide theoretical foundation and technical support for drone change detection and illegal building detection.