A method has been developed for constructing three-dimensional models of rigid objects on the earth’s surface using one satellite image using the example of railway infrastructure. The method consists in step-by-step processing of satellite images with sequential application of two convolutional neural networks. In the first processing step, a satellite image is segmented by a neural network to select a plurality of objects of predetermined classes. At the second stage of processing with the help of neural network local analysis of image areas detected by results of the first stage of processing is performed. The results of the second processing step are used to estimate the parameters of the 3D model of the object. The possibilities of the method are shown by the example of processing a satellite image of the railway infrastructure. The following classes of informative areas are considered: building, wall edge, roof edge, building shadow, railway infrastructure, car, highway; rails, poles and shadows from poles (taken as reference objects for estimating scaling coefficients in certain directions). An example is given of using the developed method of highlighting typical railway infrastructure objects and for subsequent evaluation of the parameters of a three-dimensional building model partially obscured by trees.
We proposed an approach for estimating the shape and geometric parameters of the observed objects from a perspective image based on typed elements, perspective geometry methods and convolutional neural networks. The proposed method uses the assumption that the object under study is rigid. A method is proposed for restoring a 3-D model of an observed object from one perspective image using reference objects and typed elements. Semantic segmentation of typed elements allows to set the photometric parameters of the coordinate system attached to the points on the image. According to the calculated photometric parameters and segmentation of the observed object in the image, its parameters and a 3-D model are estimated. The developed method is applicable for calculating 3-D models from a single perspective image in the vicinity of a road (both road and railway) infrastructure, where there are a large number of typed elements.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.