Ralf Reulke

Combinatorial Image Analysis

This volume presents the proccedings of the 11th International Workshop on Combinatorial Image Analysis. IWCIA 2006 was the 11th in a series of international workshopfs devoted to combinatorial image analysis. Prior meetings took place in Paris (France 1991), Ube (Japan 1992), Wahington DC (USA 1994), Lyon (France 1995), Hiroshima (Japan 1997), Madras (India 1999), Philadelphia (USA 2001), Palermo (Italy 2003) and Auckland (New Zealand 2004). For this workshop we received 59 papers from all over the world. Each paper was assigned to three independent referees and carefully revised. Finally, we selected 34 papers for the conference based on content, significance, relevance, and presentation. Conference papers are presented in this volume in the order they were presented at the conference. The topics of the conference covered combinatorial image analysis, grammars and models for analysis and recognition of scenes or images, combinatorial topology and geometry for images, digital geometry of curves or surfaces, algebraic approaches to image processing, image, point-clouds or surface registration as well as fuzzy and probabilistic image analysis. The programm followed a single-track format with presentations of all published conference papers. Non-overlapping oral and poster sessions ensured that all attendees had opportunities to interact personny with presenters. Among the highlights of the meeting were the talks of our two invited speakers, renowned experts in the field of descrete geometry, digital topology, and image analysis. - David Coeurjolly (University of Lyon, France): Computational Aspects of Digital Plane and Hyperplane Recognition - Longin Jan Latecki (Temple University, Philadelphia, USA): Polygonal Approximation of Point Sets.

SENSOR: a tool for the simulation of hyperspectral remote sensing systems

Abstract The consistent end-to-end simulation of airborne and spaceborne earth remote sensing systems is an important task, and sometimes the only way for the adaptation and optimisation of a sensor and its observation conditions, the choice and test of algorithms for data processing, error estimation and the evaluation of the capabilities of the whole sensor system. The presented software simulator SENSOR (Software Environment for the Simulation of Optical Remote sensing systems) includes a full model of the sensor hardware, the observed scene, and the atmosphere in between. The simulator consists of three parts. The first part describes the geometrical relations between scene, sun, and the remote sensing system using a ray-tracing algorithm. The second part of the simulation environment considers the radiometry. It calculates the at-sensor radiance using a pre-calculated multidimensional lookup-table taking the atmospheric influence on the radiation into account. The third part consists of an optical and an electronic sensor model for the generation of digital images. Using SENSOR for an optimisation requires the additional application of task-specific data processing algorithms. The principle of the end-to-end-simulation approach is explained, all relevant concepts of SENSOR are discussed, and first examples of its use are given. The verification of SENSOR is demonstrated. This work is closely related to the Airborne PRISM Experiment (APEX), an airborne imaging spectrometer funded by the European Space Agency.

EYESCAN - A High Resolution Digital Panoramic Camera

A digital panoramic camera system is introduced consisting of a CCD line scanner and a high precision turntable. This combination allows the use of such a digital imaging systems for photogrammetric, robot vision, artistic and other applications. Additionally, these images with a reduced resolution or parts of it can be used for internet applications. Typical fields of application are photogrammetry in architecture, digital archiving of cultural objects and virtual reality. The imaging geometry causes panoramic distortions, therefore the images must be transformed into plane coordinate systems in order to work with them. Basic equations are given for these projections.

Mobile Panoramic Mapping Using CCD-Line Camera and Laser Scanner with Integrated Position and Orientation System

The fusion of panoramic camera data with laser scanner data is a new approach and allows the combination of high-resolution image and depth data. Application areas are city modelling, virtual reality and documentation of the cultural heritage. Panoramic recording of image data is realized by a CCD-line, which is precisely rotated around the projection centre. In the case of other possible movements, the actual position of the projection centre and the view direction has to be measured. Linear moving panoramas e.g. along a wall are an interesting extension of such rotational panoramas. Here, the instantaneous position and orientation determination can be realized with an integrated navigation system comprising differential GPS and an inertial measurement unit. This paper investigates the combination of a panoramic camera and a laser scanner with a navigation system for indoor and outdoor applications. First it will be reported about laboratory experiments, which were carried out to obtain valid parameters about the surveying accuracy achievable with both sensors panoramic camera and laser scanner respectively. Then out door surveying results using a position and orientation system as navigation sensor will be presented and discussed.

Tree species recognition with fuzzy texture parameters

The management and planning of forests presumes the availability of up-to-date information on their current state. The relevant parameters like tree species, diameter of the bowl in defined heights, tree heights and positions are usually represented by a forest inventory. In order to allow the collection of these inventory parameters, an approach aiming at the integration of a terrestrial laser scanner and a high resolution panoramic camera has been developed. The integration of these sensors provides geometric information from distance measurement and high resolution texture information from the panoramic images. In order to enable a combined evaluation, in the first processing step a co-registration of both data sets is required. Afterwards geometric quantities like position and diameter of trees can be derived from the LIDAR data, whereas texture parameters are derived from the high resolution panoramic imagery. A fuzzy approach was used to detect trees and differentiate tree species.

On performance analysis of optical flow algorithms

Literally thousands of articles on optical flow algorithms have been published in the past thirty years. Only a small subset of the suggested algorithms have been analyzed with respect to their performance. These evaluations were based on black-box tests, mainly yielding information on the average accuracy on test-sequences with ground truth. No theoretically sound justification exists on why this approach meaningfully and/or exhaustively describes the properties of optical flow algorithms. In practice, design choices are often made based on unmotivated criteria or by trial and error. This article is a position paper questioning current methods in performance analysis. Without empirical results, we discuss more rigorous and theoretically sound approaches which could enable scientists and engineers alike to make sufficiently motivated design choices for a given motion estimation task.

Situation analysis and atypical event detection with multiple cameras and multi-object tracking

This paper discusses a trajectory based recognition algorithm to expand the common approaches for atypical event detection in multi-object traffic scenes and to obtain area-based types of information (e.g. maps of speed patterns, trajectory curvatures or erratic movements). Different views of the same area by more than one camera sensor are necessary, because of the typical limitations of single camera systems, resulting from occlusions by other cars, trees and traffic signs. Furthermore, distributed cooperative multi-camera system (MCS) enables a significant enlargement of the observation area. The fusion of object data from different cameras is done by a multi-target tracking approach. This approach opens up opportunities to identify and specify traffic objects, their location, speed and other characteristic object information. New and consolidated information of traffic participants is derived from the system.

A sensor-based approach to image quality

In the past 20 years a large effort has been made to characterize the image quality of remote sensing systems. The image quality can actually be measured only by the quality of the final product (e.g. object detection, classification). One option now is to use the National Image Interpretability Rating Scales (NIIRS), because NIIRS is related to object detection. From an engineering standpoint a task-based scale, like NIIRS is not well suited, because it cannot be derived from the fundamental sensor and scene behaviour. Therefore, the aim of this paper is to derive an image quality criterion, based on the physical characteristics of sensor and scene. To assess the image quality, we compare the output of the real sensor with the output of an ideal sensor based on a Local Mean Squared Error (LMSE). This criterion, we abbreviate in the following with YAIQC (Yet Another Image Quality Criteria)

SNR Evaluation of the RapidEye Space-borne Cameras

Summary: After launch and continuous radiation exposure, space-borne cameras are constantly changing. Therefore permanent technical specification and evaluation of the sensor in space plays an important role in the remote sensing community. There are a variety of evaluation criteria, which are all based on the essential camera parameters – the spatial resolution, point spread function (PSF) and noise. Noise estimation is a challenging task for characterization of remote sensing systems in space. The in-flight measurement of noise will often be done with artificial test sites. If these test sites are not suffi-ciently available, homogeneous image regions (desert, snow, water surfaces) are often used. The al-bedo of these objects, however, lies normally outside the specified albedo range of remote sensing systems focused on the Earths surface. The only possibility to determine the noise after the satellite launch within the normal operational albedo range is to use normal surface objects within the oper-ationally acquired imagery. As these objects have to be homogeneous, one needs methods that can detect the smallest homogeneous areas in the image to evaluate noise. In this paper an approach for determining the signal to noise ratio (SNR) with data from natural tar-gets is presented. In experiments, the results demonstrate that the described method performs well and results are comparable to the standard methods used to determine SNR.

Simulation of APEX data: the SENSOR approach

The consistent simulation of airborne and spaceborne hyperspectral data is an important task and sometimes the only way for the adaptation and optimization of a sensor and its observing conditions, the choice and test of algorithms for data processing, error estimations and the evaluation of the capabilities of the whole sensor system. The integration of three approaches is suggested for the data simulation of APEX (Airborne Prism Experiment): (1) a spectrally consistent approach (e.g. using AVIRIS data), (2) a geometrically consistent approach (e.g. using CASI data), and (3) an end-to- end simulation of the sensor system. In this paper, the last approach is discussed in detail. Such a technique should be used if there is no simple deterministic relation between input and output parameters. The simulation environment SENSOR (Software Environment for the Simulation of Optical Remote Sensing Systems) presented here includes a full model of the sensor system, the observed object and the atmosphere. The simulator consists of three parts. The first part describes the geometrical relations between object, sun, and sensor using a ray tracing algorithm. The second part of the simulation environment considers the radiometry. It calculates the at-sensor-radiance using a pre-calculated multidimensional lookup-table for the atmospheric boundary conditions and bi- directional reflectances. Part three consists of an optical and an electronic sensor model for the generation of digital images. Application-specific algorithms for data processing must be considered additionally. The benefit of using an end- to-end simulation approach is demonstrated, an example of a simulated APEX data cube is given, and preliminary steps of evaluation of SENSOR are carried out.