Window Detection in Facades

This work is about a novel methodology for window detection in urban environments and its multiple use in vision system applications. The presented method for window detection includes appropriate early image processing, provides a multi-scale Haar wavelet representation for the determination of image tiles which is then fed into a cascaded classifier for the task of window detection. The classifier is learned from a Gentle Adaboost driven cascaded decision tree [1] on masked information from training imagery and is tested towards window based ground truth information which is -together with the original building image databases -publicly available [9, 10, 12]. The experimental results demonstrate that single window detection is to a sufficient degree successful, e.g. for the purpose of building recognition, and, furthermore, that the classifier is in general capable to provide a region of interest operator for the interpretation of urban environments. The extraction of this categorical information is beneficial to index into search spaces for urban object recognition as well as aiming towards providing a semantic focus for accurate post-processing in 3D information processing systems. Targeted applications are (i) mobile services on uncalibrated imagery, e.g. for tourist guidance, (ii) sparse 3D city modeling, and (iii) deformation analysis from high resolution imagery.

5th Int. Symp. on Mobile Mapping Technology, 2007

This work is about a novel methodology for window detection in urban environments and its multiple use in vision system applications. The presented method for window detection includes appropriate early image processing, provides a multi-scale Haar wavelet representation for the ...

ISPRS Journal of Photogrammetry and Remote Sensing, 2014

This paper deals with the identification of blind building façades, i.e. façades which have no openings, in wide angle aerial images with a decimeter pixel size, acquired by nadir looking cameras. This blindness characterization is in general crucial for real estate estimation and has, at least in France, a particular importance on the evaluation of legal permission of constructing on a parcel due to local urban planning schemes. We assume that we have at our disposal an aerial survey with a relatively high stereo overlap alongtrack and across-track and a 3D city model of LoD 1, that can have been generated with the input images. The 3D model is textured with the aerial imagery by taking into account the 3D occlusions and by selecting for each façade the best available resolution texture seeing the whole façade. We then parse all 3D façades textures by looking for evidence of openings (windows or doors). This evidence is characterized by a comprehensive set of basic radiometric and geometrical features. The blindness prognostic is then elaborated through an (SVM) supervised classification. Despite the relatively low resolution of the images, we reach a classification accuracy of around 85% on decimeter resolution imagery with 60 Â 40% stereo overlap. On the one hand, we show that the results are very sensitive to the texturing resampling process and to vegetation presence on façade textures. On the other hand, the most relevant features for our classification framework are related to texture uniformity and horizontal aspect and to the maximal contrast of the opening detections. We conclude that standard aerial imagery used to build 3D city models can also be exploited to some extent and at no additional cost for facade blindness characterisation.

2011

Abstract We present an automatic approach to window and façade detection from LiDAR (Light Detection And Ranging) data collected from a moving vehicle along streets in urban environments. The proposed method combines bottom-up with top-down strategies to extract façade planes from noisy LiDAR point clouds. The window detection is achieved through a two-step approach: potential window point detection and window localization.

1999

In this paper we presented a novel approach for building detection from multiple aerial images in dense urban areas. The approach is based on accurate surface reconstruction, followed by extraction of visible building facades, then used as a main cue for building detection. For the facade detection a simple but nevertheless flexible and robust algorithm is proposed. It is based on observation that building facades correspond to the accumulation of 3D data, available from different views, in the object space. Knowledge-driven thresholding of 3D data accumulators followed by Hough transformbased segment detection results in facade extraction. The extracted facades are not only explicitly geo-referenced, but their visible sides are also tracked. Then, 3D regions resulting from surface reconstruction procedure and bounded by the extracted facades are detected as building hypotheses through testing a set of spatial criteria. The approach has been tested on the complex and dense urban scenes, and demonstrated a good rate of building detection.