Abstract
Abstract. For extraction of road pixels from combined image and elevation data, Wegner et al. (2015) proposed classification of superpixels into road and non-road, after which a refinement of the classification results using minimum cost paths and non-local optimization methods took place. We believed that the variable set used for classification was to a certain extent suboptimal, because many variables were redundant while several features known as useful in Photogrammetry and Remote Sensing are missed. This motivated us to implement a variable selection approach which builds a model for classification using portions of training data and subsets of features, evaluates this model, updates the feature set, and terminates when a stopping criterion is satisfied. The choice of classifier is flexible; however, we tested the approach with Logistic Regression and Random Forests, and taylored the evaluation module to the chosen classifier. To guarantee a fair comparison, we kept the segment-based approach and most of the variables from the related work, but we extended them by additional, mostly higher-level features. Applying these superior features, removing the redundant ones, as well as using more accurately acquired 3D data allowed to keep stable or even to reduce the misclassification error in a challenging dataset.
Highlights
Motivation Extracting roads from remote sensing images is important in a number of different applications, for example traffic management, city planning, road monitoring, GPS navigation and map updating (Wang et al, 2016)
Evaluation strategy The models we compare in this work use feature sets resulting from: For convenience, we summarize below our three-step algorithm: 1. Perform variable selection on a part of the training data which yields a feature subset
We found discrimination thresholds for both of our classification methods (0.5 for random forests and 0.65 for logistic regression) that result in improvements in all three of those values at the same time
Summary
Motivation Extracting roads from remote sensing images is important in a number of different applications, for example traffic management, city planning, road monitoring, GPS navigation and map updating (Wang et al, 2016). Until now, no fully automated road network detection method is applied in praxis, see Rottensteiner et al (2013). The reason lies in the difficulty of the classification problem. There are many factors that make road network extraction from remote sensing images challenging. The dependence on the sensors and resolution cause a variety of problems; even if same sensors and resolutions are used, appearance of roads in remote sensing images can vary wildly. Some reasons are weather and illumination effects and, very importantly, shadows and occlusions caused by high buildings, tree crowns, moving or parking vehicles and tiles of rubbish in developing countries. Because buildings and vehicles are indispensable part of urban terrain, road extraction from urban scenes is considered even more difficult than in rural areas (Hu et al, 2014)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have