This study proposes a hierarchical classification method using the OBIA technique for urban object detection on multi-view aerial imagery. We classify the images into seven classes, i.e. the Grass, Tree, Façade, Roof, Road, Window and Others. In the classification flow path, we utilize spectrum, geometry, class related features and the auxiliary gradient map and height map to detect different classes. Through these supplementary data, we can easily separate the surface objects with different height and get classification result with high percent correctness. In the image segmentation and region grow steps, we use the edge map and enhanced image to avoid incorrect segmentation.
Besides, we set distinct scales for detecting the surface objects in different sizes in segmentation. Then, on account of the oblique imaging geometry of OAI in varied spatial resolution, we separate the image into three parts along the viewing direction and set three gradual scales in region grow. The classification results are concluded in the following:
(1) In accuracy analyses, the overall accuracy of OAI can achieve 81% and the kappa index is 0.75 with five classes include Grass, Tree, Façade, Roof and Road. The VAI is also quantified that a 77% overall accuracy and 0.7 kappa index can be obtained. According to the user accuracy and producer accuracy of each class, we found they all can reach a certain level. It proves that the proposed classification technique is considerably effective and reliable.
(2) The problems when generating the auxiliary information are the factors that lead the wrong classification result. In height map generation, since the utilized DEM is not generated from the image matching point cloud that may induced some errors and caused part of ground miss-classified as the roofs. Holes and boundary problems in height map interpolation also influence the accuracy of classification. For these reasons, we may acquire better classification result based on the right DEM and finer the interpolation in the vertical direction on the height map. On the other hand, in gradient map generation, point cloud segmentation parameters are the decisive factors which affect the façade detection result. In our case, we tested two sets of split and merge distances and found out that the lower
67
buildings need smaller distance in order to distinguish different planes successfully within the noisy photogrammetric point cloud. With the development of the dense matching technique, the problems such as noise and outliers could be handled in the future for acquiring more complete and accurate points.
(3) The classification result includes seven classes namely Façade, Roof, Grass, Tree, Road, Window and Others. Tree and Roof have the higher accuracy both in the OAI
and VAI. Road detection is unstable since its variable size and shape affected by shadow and trees along the road. Façade on individual or large buildings are easier to be detected but could be missed in low residential area. Shadow is a critical factor to interfere classification result especially in the city area which induced incorrect image segmentation result.
Window with single color and rectangular shape can further be recognized from the
detected façade class but failed in complicated structural facades. Nevertheless, the classification consistence is high for images acquired from different view angles at the same area. It again confirms that the high reliability of the proposed urban object detection scheme.
(4) The image classification result is also projected on the photogrammetric point cloud in our research that makes the users obtain more useful materials. For all that, those semantic classes in image classification and point classification results are satisfied on land cover survey, 3D GIS construction and 3D city model reconstruction.
68
REFERENCES
Addink, E. A., Van Coillie, F. M. B., de Jong, S. M., 2012. Introduction to the GEOBIA 2010 special issue: From pixels to geographic objects in remote sensing image analysis.
International Journal of Applied Earth Observation and Geoinformation, 15, pp. 1–6.
Baatz, M., Schäpe, M., 2000. Multiresolution segmentation: An optimization approach for high quality multi-scale image segmentation. Angewandte Geographische Informations-Verarbeitung XII, Beiträge zum AGIT-Symposium Salzburg 2000 (Karlsruhe: Herbert Wichmann Verlag), pp. 12–23.
Blaschke, T., 2010. Object based image analysis for remote sensing. ISPRS Journal of Photogrammetry and Remote Sensing, 65 (1), pp. 2-16.
Canny, J., 1986. A computational approach to edge detection. IEEE Trans. Pattern Analysis Machine Intelligence 8, pp. 679–714 .
Chen, Y., Su, W., Li, J., Sun, Z., 2009. Hierarchical object oriented classification using very high resolution imagery and LIDAR data over urban areas. Advances in Space Research, 43(7), pp. 1101–1110.
Cheng, L., Tong, L., Chen, Y., Zhang, W., Shan, J., Liu, Y., Li, M., 2013. Integration of LiDAR data and optical multi-view images for 3D reconstruction of building roofs.
Optics and Lasers in Engineering, 51 (4), pp. 493–502.
Congalton, R.G., 1991. A review of assessing the accuracy of classifications of remotely sensed data. Remote Sensing of Environment, 37 (1), 35–46.
Davis, J. C., 2002. Statistics and data analysis in geology, Kansas Geological Survey. The University of Kansas, Kansas.
Frueh, C. , Sammon, R., Zakhor, A., 2004. Automated texture mapping of 3d city models with oblique aerial imagery. 3DPVT '04 Proceedings of the 3D Data Processing, Visualization, and Transmission, 2nd International Symposium, pp. 396-403.
Gerke, M., 2009. Dense matching in high resolution oblique airborne images. CMRT09:
Object extraction for 3D city models, road databases and traffic monitoring: concepts,
69
algorithms and evaluation, Paris, 3-4 September 2009. ISPRS, 2009. pp 77-82.
Gerke, M., Nyaruhuma, A. P., 2009. Incorporating scene constraints into the triangulation of airborne oblique images. Presented at the ISPRS conference: High-Resolution Earth Imaging for Geospatial Information: ISPRS XXXVIII 1-4-7/WS, 2-5 June, 2009 Hannover, Germany: 6.
Gerke. M., Kerle. N., 2011. Automatic structural seismic damage assessment with airborne oblique pictometry imagery. Photogrammetric Engineering and Remote Sensing, 77 (9), pp. 885-898.
Gröger, G., Kolbe, T. H., Nagel, C., Häfele, K. H., 2012. OGC City Geography Markup Language (CityGML) Encoding Standard, Version 2.0.0. Open Geospatial Consortium, OGC Doc. No. 12–019.
Gruen, A., 2012. Development And Status Of Image Matching In Photogrammetry. The Photogrammetric Record, 27(137), pp. 36–57.
Gruen, A., Baltsavias, E. P., 1988. Geometrically constrained multiphoto matching.
Photogrammetric Engineering & Remote Sensing, 54(5), pp. 633–641.
Haala, N., Kada, M., 2010. An update on automatic 3D building reconstruction. ISPRS Journal of Photogrammetry and Remote Sensing, 65 (6), pp. 570–580.
Haralick, R. M., Shanmugam K., Dinstein J., 1973. Textural features for image classification. IEEE Trans. Syst., Man, Cybern., vol. SMC-3, pp.610 -621.
Kao, D. L., Kramer, M. G., Love, A. L., Dungan, J. L., Pang, A. T., 2005. Visualizing distributions from multi-return LiDAR data to understand forest structure. The Cartographic Journal, 42, No. 1, pp. 35–47.
Kost, K., Loddenkemper, M., Petring, J., 1996. Airborne laserscanning, a new remote sensing method for mapping terrain. Third EARSeL Workshop on LiDAR remote sensing of land and sea, Tallinne, Estonia, 17-19 July 1997, pp. 89-96.
LI, M., 1990. High-precision relative orientation using feature based matching techniques.
ISPRS Journal of Photogrammetry and Remote Sensing, 44, pp. 311-324.
70
Lang, S., Albrecht, F., Blaschke, T., 2011. OBIA – Tutorial, Centre for Geoinformatics (Z_GIS), Paris-Lodron University Salzburg.
Megyesi, Z., 2009. Dense Matching Methods for 3D Scene Reconstruction from Wide Baseline Images. PhD Dissertation, Computer and Automation Research Institute, Eötvös Loránd University, Budapest.
Mishra, P., Ofek, E., Kimchi, G., 2008. Validation of vector data using oblique images.
Proceedings of the 16th ACM SIGSPATIAL international conference on Advances in geographic information systems, November 05-07, 2008, Irvine, California.
Nyaruhuma, A. P., Gerke, M., Vosselman, G., 2010. Line matching in oblique airborne images to support automatic verification of building outlines, ASPRS 2010 Annual Conference.
Panday, U. S., 2011. Fitting of parametric building models to oblique aerial images, Faculty of Geo-Information Science and Earth Observation, University of Twente.
Sheng, Y., Gong, P., Biging, G.S., 2003. True orthoimage production for forested areas from large-scale aerial photographs. Photogrammetric Engineering and Remote Sensing, 69, pp. 259–266.
Strecha, C., Tuytelaars, T., Van Gool, L., 2003. Dense matching of multiple wide-baseline views. Ninth IEEE International Conference on Computer Vision, 2 , pp. 1194-1201.
Tola, E., Lepetit, V., Fua, P., 2010. DAISY: An efficient dense descriptor applied to wide-baseline stereo, IEEE Transactions On Pattern Analysis And Machine Intelligence, 32, No. 5, pp 815-830.
Trimble. 2010. eCognition Developer 8.0.2 - Reference Book. Munchen,Germany: Trimble Germany GmbH.
Vallet, J., Panissod , F., Strecha, C., Tracol M., 2011. Photogrammetric performance of an ultra light weight swinglet“UAV”. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, ISPRS ICWG I/V UAV-g (unmanned aerial vehicle in geomatics) conference, Zurich, Switzerland.
71
Wang, L., Neumann, U., 2009. A robust approach for automatic registration of aerial images with untextured aerial LiDAR data. Computer Vision and Pattern Recognition, IEEE Conference on, pp. 2623 – 2630.
Wang, M., Tseng, Y. H., 2011. Incremental segmentation of LiDAR point clouds with an octree-structured voxel space. The Photogrammetric Record, 26(133), pp. 32–57.
Wolf, P. R., Dewitt, B. A., 2000. Elements of photogrammetry. McGraw-Hill, New York, USA.
Xiao, J., Gerke, M. and Vosselman, G., 2012. Building extraction from oblique airborne imagery based on robust façade detection. ISPRS Journal of Photogrammetry and Remote Sensing, 68, pp. 56-68.
Zebedin, L., Klaus, A., Gruber-Geymayer, B., Karner, K., 2006, Towards 3D map generation from digital aerial images. ISPRS Journal of Photogrammetry and Remote Sensing, 60 (6), pp. 413–427.
Zhang, L., Gruen, A., 2006. Multi-image matching for DSM generation from IKONOS imagery. ISPRS Journal of Photogrammetry and Remote Sensing, 60 (3), pp. 195–211.