Automatic reconstruction of regular buildings using a shape‐based balloon snake model

Buildings are regarded as the most prominent man-made objects: hence, their automatic extraction from aerial and satellite images has attracted the interest of photogrammetric research communities. In this paper, a balloon snake, a well-known active contour model, was developed for the precise extraction of building boundaries. Traditional active contours are highly sensitive to image data and are problematic when building boundaries are weakened, usually due to occlusions. To improve the efficiency of the traditional balloon snake model, a shape-based balloon snake model was introduced where knowledge about the expected geometrical shape of buildings was modelled and applied to the traditional snake representation. Implementation of the proposed algorithm confirmed its efficiency in the precise extraction of building boundaries. Experimental results and the qualitative and quantitative evaluations demonstrate the potential of the proposed shape-based balloon snake model and its superiority over traditional models. Resume Les bâtiments sont consideres comme les objets artificiels les plus remarquables, c'est pourquoi leur extraction automatique a partir d'images aeriennes ou spatiales suscite l'interet de la communaute de la recherche en photogrammetrie. Cet article presente un modele de contours actifs bien connu, le «balloon snake», developpe pour l'extraction precise des limites de bâtiments. Les methodes traditionnelles de contours actifs sont tres sensibles aux donnees images et posent des problemes lorsque les limites de bâtiments n'apparaissent pas clairement, notamment en cas d'occlusion. Afin d'ameliorer les performances du «balloon snake» traditionnel, un modele base sur la forme est propose, dans lequel la forme previsible des bâtiments est modelisee et appliquee au modele «balloon snake» traditionnel. La mise en œuvre de l'algorithme propose confirme son efficacite pour l'extraction precise de contours de bâtiments. Les resultats experimentaux et leur evaluation qualitative et quantitative montrent le potentiel du modele «balloon snake» base sur la forme et sa superiorite par rapport aux modeles traditionnels. Zusammenfassung Gebaude werden als wichtigste kunstliche Objekte betrachtet, weshalb ihre automatisierte Extraktion aus Luft- und Satellitenbilddaten groses Interesse in der photogrammetrischen Forschung hervorgerufen hat. Dieser Beitrag nutzt sogenannte “Balloon-Snakes”, eine bekannte Art von Aktiven Konturmodellen, zur prazisen Erfassung von Gebaudeumrissen. Traditionelle Aktive Konturmodelle reagieren sehr sensitiv auf Bilddaten und haben grose Probleme, wenn die Gebaudeumrisse, etwa durch Verdeckungen, unklar sind. Das traditionelle Modell der “Balloon-Snakes” wird durch ein “Formbasiertes Balloon-Snakes” Modell ersetzt, bei dem die erwarteten geometrischen Formen von Gebauden modelliert werden. Experimentelle Ergebnisse und deren qualitative und quantitative Analyse bestatigen das Potential dieses “Formbasierten Balloon-Snakes” Modells und die Uberlegenheit gegenuber herkommlichen Modellen. Resumen Los edificios son considerados como los objetos mas destacados hechos por el hombre: por lo tanto, su extraccion automatica a partir de imagenes aereas y de satelite ha atraido el interes de las comunidades de investigacion fotogrametricas. En este trabajo un modelo de contorno activo bien conocido (balloon snake) fue desarrollado para la extraccion precisa de los limites de construccion. Contornos activos tradicionales son altamente sensibles a los datos de imagen y son problematicos cuando los limites de construccion se debilitan, normalmente debido a oclusiones. Para mejorar la eficiencia del modelo tradicional, un modelo basado en la forma se introdujo en el que se modela la conocida forma geometrica prevista de edificios y se aplica al modelo tradicional. La aplicacion del algoritmo propuesto confirmo su eficiencia en la extraccion precisa de los limites de construccion. Los resultados experimentales y las evaluaciones cualitativas y cuantitativas demuestran el potencial del modelo propuesto basado en la forma y su superioridad respecto a los modelos tradicionales. 摘要 从航空或者卫星影像上自动提取建筑物一直以来是摄影测量领域的热门研究领域。本文通过发展主动轮廓模型中气球蛇模型,来精确提取建筑物的边界。经典的主动轮廓提取对影像数据非常敏感,在弱房屋边界尤其是房屋边缘被遮挡的时候,一般都会得到错误的结果。本文提出基于几何形状的约束的气球蛇模型,同时利用房屋的几何模型信息和主动轮廓模型的优势,提高房屋提取的有效性。通过定性和定量的试验验证了本文提出的基于形状的气球蛇模型在建筑物自动提取上优于经典的气球蛇模型。

[1]  Demetri Terzopoulos,et al.  Snakes: Active contour models , 2004, International Journal of Computer Vision.

[2]  Faouzi Ghorbel,et al.  Fourier-based geometric shape prior for snakes , 2008, Pattern Recognit. Lett..

[3]  Robert Hooke,et al.  `` Direct Search'' Solution of Numerical and Statistical Problems , 1961, JACM.

[4]  Ilkay Ulusoy,et al.  Building detection in high resolution remotely sensed images based on morphological operators , 2009, 2009 4th International Conference on Recent Advances in Space Technologies.

[5]  Cem Ünsalan,et al.  A Probabilistic Framework to Detect Buildings in Aerial and Satellite Images , 2011, IEEE Transactions on Geoscience and Remote Sensing.

[6]  Issam Dagher,et al.  WaterBalloons: A hybrid watershed Balloon Snake segmentation , 2008, Image Vis. Comput..

[7]  Chengsu Ouyang,et al.  A Novel Snake Model for X-Ray Image Segmentation , 2009, 2009 2nd International Congress on Image and Signal Processing.

[8]  Laurent D. Cohen,et al.  On active contour models and balloons , 1991, CVGIP Image Underst..

[9]  Scott T. Acton,et al.  Affine and projective active contour models , 2007, Pattern Recognit..

[10]  Peter Reinartz,et al.  Detecting complex building shapes in panchromatic satellite images for digital elevation model enhancement , 2010 .

[11]  Hamid Abrishami Moghaddam,et al.  Automatic urban building boundary extraction from high resolution aerial images using an innovative model of active contours , 2010, Int. J. Appl. Earth Obs. Geoinformation.

[12]  Sebastien Ourselin,et al.  A New Deformable Model Using Dynamic Gradient Vector Flow and Adaptive Balloon Forces , 2003 .

[13]  Alan S. Willsky,et al.  Nonparametric shape priors for active contour-based image segmentation , 2005, 2005 13th European Signal Processing Conference.

[14]  S. Y. Lam,et al.  Enhanced Snake algorithm by embedded domain transformation , 2006, Pattern Recognit..

[15]  Matthieu Cord,et al.  Building Detection and Reconstruction from Mid- and High-Resolution Aerial Imagery , 1998, Comput. Vis. Image Underst..

[16]  Parvaneh Saeedi,et al.  Automatic building detection in aerial and satellite images , 2008, 2008 10th International Conference on Control, Automation, Robotics and Vision.

[17]  Vicent Caselles,et al.  Geometric models for active contours , 1995, Proceedings., International Conference on Image Processing.

[18]  Claus Brenner,et al.  Extraction of buildings and trees in urban environments , 1999 .

[19]  Jerry L. Prince,et al.  Snakes, shapes, and gradient vector flow , 1998, IEEE Trans. Image Process..

[20]  Nikos Paragios,et al.  Recognition-Driven Two-Dimensional Competing Priors Toward Automatic and Accurate Building Detection , 2009, IEEE Transactions on Geoscience and Remote Sensing.

[21]  Timothy L. Haithcoat,et al.  Development of comprehensive accuracy assessment indexes for building footprint extraction , 2005, IEEE Transactions on Geoscience and Remote Sensing.

[22]  You Hongjian,et al.  3D building reconstruction from aerial CCD image and sparse laser sample data , 2006 .

[23]  Horst Steuer Height snakes: 3D building reconstruction from aerial image and laser scanner data , 2011, 2011 Joint Urban Remote Sensing Event.

[24]  Heinz Rüther,et al.  Application of snakes and dynamic programming optimisation technique in modeling of buildings in informal settlement areas , 2002 .

[25]  Tee-Ann Teo,et al.  Fusion of lidar data and optical imagery for building modeling , 2004 .

[26]  Farhad Samadzadegan,et al.  Automatic 3D object recognition and reconstruction based on neuro-fuzzy modelling , 2005 .

[27]  Luc Van Gool,et al.  A probabilistic approach to roof extraction and reconstruction , 2002 .

[28]  Clive S. Fraser,et al.  Improved Building Detection Using Texture Information , 2013 .

[29]  Kourosh Khoshelham,et al.  A Model‐Based Approach to Semi‐Automated Reconstruction of Buildings from Aerial Images , 2004 .

[30]  Salman Ahmadi,et al.  An improved snake model for automatic extraction of buildings from urban aerial images and LiDAR data , 2010, Comput. Environ. Urban Syst..

[31]  Yun Zhang,et al.  Semi-Automatic Building Extraction Utilizing QuickBird Imagery , 2005 .

[32]  Andrew Zisserman,et al.  A PLANE-SWEEP STRATEGY FOR THE 3D RECONSTRUCTION OF BUILDINGS FROM MULTIPLE IMAGES , 2000 .