Scene understanding for autonomous vehicles is a challenging computer vision task, with recent advances in convolutional neural networks (CNNs) achieving results that notably surpass prior traditional feature driven approaches. However, limited work investigates the application of such methods either within the highly unstructured off-road environment or to RGBD input data. In this work, we tak...

The geographic information system industry would benefit from flexible automated systems capable of extracting linear structures from satellite imagery. Quadratic snakes allow global interactions between points along a contour, and are well suited to segmentation of linear structures such as roads. However, a single quadratic snake is unable to extract disconnected road networks and enclosed re...

Road surface extraction from remote sensing images using deep learning methods has achieved good performance, while most of the existing are based on fully supervised learning, which requires a large amount training data with laborious per-pixel annotation. In this article, we propose scribble-based weakly road method named ScRoadExtractor, learns easily accessible scribbles such as centerlines...

The problem of road segmentation in high resolution images is addressed in this paper. We present an extension of the active contour model that includes a-priori knowledge about the width of the roads being extracted. In order to improve the segmentation performance of the algorithm, the proposed model also contains a modified external energy term. The problem of the contour energy minimization...

sar (synthetic aperture radar) image enhancement and segmentation is purpose of this thesis. sar image segmentation is a primary step before steps such as classification and target recognition. the main obstacle in sar image segmentation is inherent speckle noise. speckle noise is a multiplicative and highly destructive noise which results to intensity inhomogeneity. hence common segmentation m...

Dynamic segmentation is a method that facilitates the small areas along a line feature to be referenced without actually breaking the line into pieces. However, the current dynamic segmentation methods are almost developed from 2D roadway centerline network models, and therefore they are unable to support 3D lane-oriented inventory management (e.g. pavement condition, traffic volumes, traffic a...

The field of deep learning has seen significant advancement in recent years. However, much of the existing work has been focused on real-valued numbers. Recent work has shown that a deep learning system using the complex numbers can be deeper for a fixed parameter budget compared to its real-valued counterpart. In this work, we explore the benefits of generalizing one step further into the hype...

An approach for the semi-automatic extraction of roads from high-resolution satellite imagery is proposed. Scale space, Edge-detection techniques are used as pre-processing for segmentation and estimation of road width. The detection of road is based on the cost minimization technique. The cost is estimated by taking various factors into consideration like variance, direction, length and width ...

This paper explores and investigates Deep Convolutional Neural Networks (DCNNs) architectures to increase efficiency and robustness of semantic segmentation tasks. The proposed solutions are based on Up-Convolutional Networks. We introduce three different architectures in this work. The first architecture, called Part-Net, is designed to tackle the specific problem of human body part segmentati...

Construction, development and maintenance of the road network are central activities for several public authorities. In cooperation with the Norwegian road authorities, we have developed an approach for automated vehicle detection and generation of traffic statistics from QuickBird images. Satellite surveillance serves several obvious advantages over the methods that are being used today, which...