Extended Phase Field Higher-Order Active Contour Models for Networks

Abstract

This paper addresses the segmentation from an image of entities that have the form of a 'network', i.e. the region in the image corresponding to the entity is composed of branches joining together at junctions, e.g. road or vascular networks. We present new phase field higher-order active contour (HOAC) prior models for network regions, and apply them to the segmentation of road networks from very high resolution satellite images. This is a hard problem for two reasons. First, the images are complex, with much 'noise' in the road region due to cars, road markings, etc., while the background is very varied, containing many features that are locally similar to roads. Second, network regions are complex to model, because they may have arbitrary topology. In particular, we address a limitation of a previous model in which network branch width was constrained to be similar to maximum network branch radius of curvature, thereby providing a poor model of networks with straight narrow branches or highly curved, wide branches. We solve this problem by introducing first an additional nonlinear nonlocal HOAC term, and then an additional linear nonlocal HOAC term to improve the computational speed. Both terms allow separate control of branch width and branch curvature, and furnish better prolongation for the same width, but the linear term has several advantages: it is more efficient, and it is able to model multiple widths simultaneously. To cope with the difficulty of parameter selection for these models, we perform a stability analysis of a long bar with a given width, and hence show how to choose the parameters of the energy functions. After adding a likelihood energy, we use both models to extract the road network quasi-automatically from pieces of a QuickBird image, and compare the results to other models in the literature. The state-of-the-art results obtained demonstrate the superiority of our new models, the importance of strong prior knowledge in general, and of the new terms in particular.

Recovering Line-Networks in Images by Junction-Point Processes

“…extraction [16]. Several methods exploiting geodesic distances have also been proposed, as [15] where tubular structures are extracted from bi-dimensional images by computing geodesic curves over a four-dimensional space that includes local orientation and scale.…”

“…The accuracy of our algorithm has been evaluated on the road extraction problem from satellite images, and compared to existing methods [16,21,22]. The input image shown on Fig.…”

“…Time Junction-points 0.825 0.769 graph 25min Peng et al [16] 0.852 0.969 pixel 60min Yu et al [22] 0.369 0.605 line n/a Wang et al [21] 0.346 0.935 pixel n/a Figure 10. Comparison with three specialized algorithms dedicated to road-network extraction [16,21,22].…”

Morphological road segmentation in urban areas from high resolution satellite images

“…In former works based on mid-resolution sources, the solution has been often limited to the detection of non-urban or main urban roads, like in [5], while in more recent times the attention moved on high resolution imagery, with the development of several automatic and semiautomatic techniques for urban roads detection [6,7,8,9]. Different methodologies and approaches have been proposed, from tracking based methods [10], where the detection process relies on the generation of seeds that help following the road structures along the images, techniques based on shape analysis [11], to the methods based on higher order active contours, like in [12,13], that typically need a higher computational effort. To not forget are also general segmentation techniques that whether properly defined can provide road extraction as a by-product, as is the case of recent texture-based segmentation techniques [14,15] Since many of these methods, but the works of Peng's et al [12,13], above all those aiming at the automatic extraction of road information, has proved their potentials only in simple scenarios, and are likely to fail in complex environments like dense urban areas, in this paper we decided to take back into consideration the use of morphological analysis to cope with this problem, and present here a technique of road segmentation that completely relies on such image processing domain.…”

“…Different methodologies and approaches have been proposed, from tracking based methods [10], where the detection process relies on the generation of seeds that help following the road structures along the images, techniques based on shape analysis [11], to the methods based on higher order active contours, like in [12,13], that typically need a higher computational effort. To not forget are also general segmentation techniques that whether properly defined can provide road extraction as a by-product, as is the case of recent texture-based segmentation techniques [14,15] Since many of these methods, but the works of Peng's et al [12,13], above all those aiming at the automatic extraction of road information, has proved their potentials only in simple scenarios, and are likely to fail in complex environments like dense urban areas, in this paper we decided to take back into consideration the use of morphological analysis to cope with this problem, and present here a technique of road segmentation that completely relies on such image processing domain. This choice is justified by the fact that the current resolutions provide data with a considerable richness in fine details, that implicitly enhances the geometric properties of the detected scenes.…”