Detecting Building Occlusion From Aerial “ghost” Imagery

Compared with other remote sensing methods,aerial imagery has the advantages of large field of view,high timeliness,high resolution and high accuracy of model matching.Therefore,aerial images are used to make true images of urban images most of time.However,due to the majority of aerial imagery using the central projection of the imaging method,the projection method is easy to produce projection difference,and the building is a high degree of artificial objects,so that in imaging,the building always caused tilt,deformation,and always occlusion ground or other buildings.Therefore,solving the problem of building occlusion surface in aerial imagery becomes the key technology of true orthoimage generation in urban area.The true orthoimage of the urban area is generated by orthophoto rectification and occlusion detection on the basis of the original aerial image.The true orthoimage has clear and unobstructed characteristics,which can accurately reflect the basic information such as location of building and road,which have great significance in national production and urban development.The mostly methods of occlusion detection are based on the original images.In the process of orthorectification by using DSM and DBM,the mostly methods are adding the influence condition of judging occlusion.However,these occlusion detection methods are always complex and the accuracy of detection are uncertain,due to the combination of the orthorectification process,making the occlusion detection process becomes complex and the accuracy can not be controlled.In according to the former study and the principle of occlusion generation,this paper proposes a new method which detects occlusion by using “ghost” image.The traditional orthorectification does not take into account the occlusion factor,after the pixel resampling process,the buildings of orthoimage is chaos and contains the wall and two roofs.Which performance of chaos image is called “ghost” image.By analyzing the causes of the formation of “ghost” images,and analyzing the relationship between the occlusion area and the original image and the “ghost” image,it is found that the cause of the occlusion is that the high-rise artificial facilities such as buildings arises projection displacement which are caused by the central projection.The projection of building occludes the ground and other objects.In “ghost” image,the occlusion area is equal to the area of double-mapping.Firstly,this paper determines the boundary of occlusion area by using the DBM and internal and external parameters of camera.Then,the region growing method can be used to detect and compensate the occlusion area.The experimental results show that the proposed method has a success rate of 99.2% for the occlusion region,which avoids the pseudo-occlusion and pseudo-visible in the previous methods.The advantage of this method is that the occlusion detection and the true orthoimage generation is synchronization,so that this method is simple and effective.