Abstract
In this paper, we consider building extraction from high spatial resolution remote sensing images. At present, most building extraction methods are based on artificial features. However, the diversity and complexity of buildings mean that building extraction methods still face great challenges, so methods based on deep learning have recently been proposed. In this paper, a building extraction framework based on a convolution neural network and edge detection algorithm is proposed. The method is called Mask R-CNN Fusion Sobel. Because of the outstanding achievement of Mask R-CNN in the field of image segmentation, this paper improves it and then applies it in remote sensing image building extraction. Our method consists of three parts. First, the convolutional neural network is used for rough location and pixel level classification, and the problem of false and missed extraction is solved by automatically discovering semantic features. Second, Sobel edge detection algorithm is used to segment building edges accurately so as to solve the problem of edge extraction and the integrity of the object of deep convolutional neural networks in semantic segmentation. Third, buildings are extracted by the fusion algorithm. We utilize the proposed framework to extract the building in high-resolution remote sensing images from Chinese satellite GF-2, and the experiments show that the average value of IOU (intersection over union) of the proposed method was 88.7% and the average value of Kappa was 87.8%, respectively. Therefore, our method can be applied to the recognition and segmentation of complex buildings and is superior to the classical method in accuracy.
Highlights
With the development of remote sensing satellite technology and the demand of urbanization, it has become an important research field to automatically and accurately extract building objects from remote sensing images
The second is the building extraction method combined with elevation information, and this kind of method uses elevation information to separate out non-ground points, and detect buildings by combining the common edge features, spectral features and other artificial features [2]
This paper investigates the possibility of artificial features to optimize the building extraction results of convolutional neural networks
Summary
With the development of remote sensing satellite technology and the demand of urbanization, it has become an important research field to automatically and accurately extract building objects from remote sensing images. A fusion method of artificial edge features and convolutional neural network recognition results is proposed to address the problem of building extraction accurately. The results show that fine tuning is the best training strategy of convolutional neural networks architecture of R-CNN is divided into three parts: firstly, 2000 candidate regions are extracted by applied to remote sensing images. Yu et al proposed a convolutional neural network remote sensing selective search; secondly, the extracted candidate regions are extracted by a multi-layer convolutional classification model based on PTL-CFS (parameter transfer learning and correlation-based feature neural network; lastly, support vector machine (SVM) and linear regression model are used to classify selection), which can accelerate the convergence speed of CNN loss function [20]. Regions so as to reduce the training time of the model
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
