Fraunhofer-Institut für Optronik, Systemtechnik und Bildauswertung IOSB

Filters

4
4
Reset filters

Settings
Now showing 1 - 4 of 4
  • Publication
    An accumulating interpreter for cognitive vision production systems
    ( 2010)
    Michaelsen, E.
    ;
    Doktorski, L.
    ;
    Luetjen, K.
    Knowledge-based recognition and analysis of high dimensional data such as aerial images often has high computational complexity. For most applications time and computational resources such as memory are limited. Therefore approximately correct interpreters with any-time capability are proposed. In this contribution a special software architecture is published, which can handle the administration of complex knowledge-based recognition and analysis in a tractable manner.
  • Publication
    Object-oriented landmark recognition for UAV-navigation
    ( 2010)
    Michaelsen, E.
    ;
    Roschkowski, D.
    ;
    Doktorski, L.
    ;
    Jaeger, K.
    ;
    Arens, M.
    Computer vision is an ever more important means for the navigation of UAVs. Here we propose a landmark recognition system looking for salient man-made infrastructure. An object-oriented structural system is preferred since it can utilize known properties of these objects such as part-of hierarchies, mutual geometric constraints of parts, generalization etc. The structure, available for use as landmark, will vary strongly with the region the UAV is supposed to navigate in. Clear object-oriented coding of the knowledge on the landmarks, their constraints, and their properties is a key to swift adaption. This contribution reports on an example: Adapting a system, designed for a central European country (Germany), for use in a more Eastern region (Turkey).
  • Publication
    Accurate single image multi-modal camera pose estimation
    ( 2010)
    Bodensteiner, C.
    ;
    Hebel, Marcus
    ;
    Arens, M.
    A well known problem in photogrammetry and computer vision is the precise and robust determination of camera poses with respect to a given 3D model. In this work we propose a novel multi-modal method for single image camera pose estimation with respect to 3D models with intensity information (e.g., LiDAR data with reectance information). We utilize a direct point based rendering approach to generate synthetic 2D views from 3D datasets in order to bridge the dimensionality gap. The proposed method then establishes 2D/2D point and local region correspondences based on a novel self-similarity distance measure. Correct correspondences are robustly identified by searching for small regions with a similar geometric relationship of local self-similarities using a Generalized Hough Transform. After backprojection of the generated features into 3D a standard Perspective-n-Points problem is solved to yield an initial camera pose. The pose is then accurately refined using an intensity based 2D/3D registration approach. An evaluation on Vis/IR 2D and airborne and terrestrial 3D datasets shows that the proposed method is applicable to a wide range of different sensor types. In addition, the approach outperforms standard global multi-modal 2D/3D registration approaches based on Mutual Information with respect to robustness and speed. Potential applications are widespread and include for instance multispectral texturing of 3D models, SLAM applications, sensor data fusion and multi-spectral camera calibration and super-resolution applications.
  • Publication
    Perceptual grouping for building recognition from satellite SAR image stacks
    ( 2010)
    Michaelsen, E.
    ;
    Soergel, U.
    ;
    Schunert, A.
    ;
    Doktorski, L.
    ;
    Jaeger, K.
    Modern high resolution satellite SAR sensors even allow analysis of building sub-structures like windows and balconies. In the amplitude data man-made objects usually appear either as salient bright lines or points embedded within dark background. The latter features may coincide also with so-called persistent scatterers (PS), whose phase history is exploited by time series analysis for 3D reconstruction and deformation monitoring. We apply principles of human gestalt perception for grouping urban objects such as entire facades. The analysis takes place both in the amplitude and the phase data. Fusion is possible at different levels, e.g: i) grouping results in one domain may focus search in the other domain, ii) both approaches allow to infer independently the 3D structure by exploiting complementary features (amplitude vs. phase), and iii) grouping as such is useful to introduce model knowledge.