194 results for Klette, Reinhard

  • Concise Computer Vision: An Introduction into theory and algorithms

    Klette, Reinhard (2014)

    Book
    The University of Auckland Library

    Concise Computer Vision provides an accessible general introduction to the essential topics in computer vision, highlighting the role of important algorithms and mathematical concepts. Classroom-tested programming exercises and review questions are also supplied at the end of each chapter.

    View record details
  • Approximated Ground Truth for Stereo and Motion Analysis on Real-World Sequences (Poster)

    Liu, Zhifeng; Klette, Reinhard (2009)

    Conference poster
    The University of Auckland Library

    Conference details: The 3rd Pacific-Rim Symposium on Image and Video Technology (PSIVT2009) Tokyo, Japan, January 13th—16th, 2009 http://psivt2009.nii.ac.jp/ We approximate ground truth for real-world stereo sequences and demonstrate its use for the performance analysis of a few selected stereo matching and optic flow techniques. Basically we assume zero roll and constant tilt of an ego-vehicle (for about 10 seconds) driving on a planar road.

    View record details
  • Inclusion of a Second-Order Prior into Semi-Global Matching. (2009)

    Hermann, Simon; Klette, Reinhard; Destenfanis, Eduardo (2009)

    Conference poster
    The University of Auckland Library

    Conference details: The 3rd Pacific-Rim Symposium on Image and Video Technology (PSIVT2009) Tokyo, Japan, January 13th—16th, 2009 http://psivt2009.nii.ac.jp/ We consider different parameter settings for SGM, suggest to include a second order prior into the smoothness term of the energy function, and propose and test a new cost function. Some preprocessing (edge images) proves to be of value for improving SGM stereo results on real-world data.

    View record details
  • Recovery Rate of Clustering Algorithms. (2009)

    Li, Fajie; Klette, Reinhard (2009)

    Conference poster
    The University of Auckland Library

    Conference details: The 3rd Pacific-Rim Symposium on Image and Video Technology (PSIVT2009) Tokyo, Japan, January 13th—16th, 2009 http://psivt2009.nii.ac.jp/ We provide a simple and general way for defining the recovery rate of clustering algorithms using a given family of old clusters for evaluating the performance when calculating a family of new clusters. The recovery rate may be calculated by using an approximate and efficient algorithm.

    View record details
  • A Variant of Adaptive Mean Shift-Based Clustering

    Li, Fajie; Klette, Reinhard (2008)

    Conference poster
    The University of Auckland Library

    Conference Details: ICONIP 2008 - 15th International Conference on Neural Information. Processing of the Asia-Pacific Neural Network Assembly November 25-28, 2008, Auckland, New Zealand We are interested in clustering sets of highly overlapping clusters. For example, given is an observed set of stars (considered to be a set of points); how to find (recover) clusters which are the contributing galaxies of the observed union of those clusters? Below we propose a modification of an adaptive mean shift-based clustering algorithm (called Algorithm 1) proposed in 2003 by B. Geogescu, I. Shimsoni and P. Meer

    View record details
  • Landmark Initialization for Unscented Kalman Filter Sensor Fusion for Monocular Camera Localization

    Hartmann, G; Huang, F; Klette, Reinhard (2013-03)

    Journal article
    The University of Auckland Library

    The determination of the pose of the imaging camera is a fundamental problem in computer vision. In the monocular case, difficulties in determining the scene scale and the limitation to bearing-only measurements increase the difficulty in estimating camera pose accurately. Many mobile phones now contain inertial measurement devices, which may lend some aid to the task of determining camera pose. In this study, by means of simulation and real-world experimentation, we explore an approach to monocular camera localization that incorporates both observations of the environment and measurements from accelerometers and gyroscopes. The unscented Kalman filter was implemented for this task. Our main contribution is a novel approach to landmark initialization in a Kalman filter; we characterize the tolerance to noise that this approach allows.

    View record details
  • Novel Backprojection Method for Monocular Head Pose Estimation

    Ju, K; Shin, Bok Suk; Klette, Reinhard (2013-07)

    Journal article
    The University of Auckland Library

    Estimating a driver’s head pose is an important task in driver-assistance systems because it can provide information about where a driver is looking, thereby giving useful cues about the status of the driver (i.e., paying proper attention, fatigued, etc.). This study proposes a system for estimating the head pose using monocular images, which includes a novel use of backprojection. The system can use a single image to estimate a driver’s head pose at a particular time stamp, or an image sequence to support the analysis of a driver’s status. Using our proposed system, we compared two previous pose estimation approaches. We introduced an approach for providing ground-truth reference data using a mannequin model. Our experimental results demonstrate that the proposed system provides relatively accurate estimations of the yaw, tilt, and roll angle. The results also show that one of the pose estimation approaches (perspective-n-point, PnP) provided a consistently better estimate compared to the other (pose from orthography and scaling with iterations, POSIT) using our proposed system.

    View record details
  • Goal: Real-Time Segmentation via Graph Cut Goal: Real-Time Segmentation via Graph Cut

    Vaudrey, Tobi; Wedel, Andreas; Rabe, Clemens; Klappstein, Jens; Klette, Reinhard (2008)

    Conference poster
    The University of Auckland Library

    Conference Details: 2008 23rd International Conference Image and Vision Computing New Zealand. IVCNZ 08. Lincoln University, Christchurch, 26-28 November 2008. http://www.lvl.co.nz/ivcnz2008/ The detection of moving objects is a crucial part of driver assistance systems. This paper tackles this issue using computer vision. Two approaches are investigated, monocular and stereoscopic. The base principals and implementational issues are discussed and detailed, high- lighting areas of concern. In both cases, the detection is based on motion analysis of individually tracked image points (optical ow). The monoc- ular approach relies solely on the optical ow, where as the stereoscopic approach also takes stereo depth information into account. In both ap- proaches the motion analysis provides a motion metric which corresponds to the likelihood that the tracked point is moving. Based on this metric the points are segmented into objects by employing the globally op- timal graph cut algorithm. These approaches are then compared and contrasted using real-world vehicle image sequences.

    View record details
  • An Approach for Evaluating Robustness of Edge Operators using Real-World Driving Scenes

    Al-Sarraf, Ali; Vaudrey, Tobi; Klette, Reinhard; Woo, Young Woon (2008)

    Conference poster
    The University of Auckland Library

    Conference Details: 2008 23rd International Conference Image and Vision Computing New Zealand Lincoln University, Christchurch, 26-28 November 2008. http://www.lvl.co.nz/ivcnz2008/ Over the past 20 years there have been many papers that compare and evaluate di erent edge operators. Most of them focus on accuracy and also do comparisons against synthetic data. This paper focuses on real-world driver assistance scenes and does a comparison based on robustness. The three edge operators compared are Sobel, Canny and the under-publicized phase-based Kovesi-Owens operator. The Kovesi- Owens operator has the distinct advantage that it uses one pre-selected set of parameters and can work across almost any type of scene, where as other operators require parameter tuning. The results from our comparison show that the Kovesi-Owens operator is the most robust of the three, and can get decent results, even under weak illumination and varying gradients in the images. Keywords: edge operators, edge robustness evaluation, Kovesi-Owens, phase operators

    View record details
  • Animated Non-Photorealistic Rendering in Multiple Styles

    Chen, T-Y; Klette, Reinhard (2013-11)

    Conference item
    The University of Auckland Library

    This paper presents an algorithm and its applications for artistic rendering of recorded video data following ideas of non-photorealistic rendering. The proposed algorithm does not only work on a variety of artistic rendering styles for static photography but it can also be applied to the creation of artistic videos. Cartoon-like and comic-like styles are the two artistic styles considered in this paper. For creating successfully an artistic video, three key challenges are addressed: temporal (or colour) consistency, stylistic flexibility, and scalability. Our work on addressing these challenges started with collecting samples of image and video data. Based on experimental results, we designed a method for video-based non-photorealistic rendering from those input data, either in cartoon-like or in comics-like style. The paper demonstrates the benefit of the designed video-based rendering framework by comparing its results with results obtained from existing Android apps.

    View record details
  • Inverse Skeletal Strokes

    Liu, D; Klette, Reinhard (2013-11)

    Conference item
    The University of Auckland Library

    The skeletal stroke method is a general brush tool which can take a straight vector artwork as “ink”. It is easy to apply, but it is limited by the requirement of straight inputs. To offer additional input options, we present inverse skeletal strokes, a method for straightening warped vector artworks. Our method takes a user stroke to help understanding the structure of an input artwork. The key-idea is finding a set of arcs which show the “directional trend” of the artwork, and map the artwork into a new version in which these arcs are straightened. We propose a measure representing the degree of parallelism between two arcs. Using this measure, we select a set of arcs from the input artwork which are approximately parallel to the given user stroke. This is a condensed representation of a user’s intention. Then we transform the user stroke with the goal to maximize the degree of parallelism to each of the selected approximately parallel arcs. At last, we parametrize the artwork with respect to the optimized stroke, and map it into a straight version.

    View record details
  • Accuracy of Trajectories Estimation in a Driver-Assistance Context

    Khan, W; Klette, Reinhard (2014)

    Conference item
    The University of Auckland Library

    Feature-point tracking for the purpose of object tracking in a driver-assistance context is not an easy task. First, to track rigid objects, feature points have to be matched frame-by-frame and then, by using disparity maps, their real-world position can be derived, from which the object velocity is estimated. Unfortunately, a feature-point matcher cannot find (reliable) matches in all frames. In fact, the performance of a matcher varies with the type of feature-point detector and descriptor used. Our comparison of different feature-point matchers gives a general impression of how descriptor performance degrades as a rigid object approaches the ego-vehicle in a collision-scenario video sequence. To handle the mismatches, we use a Kalman-filter-based tracker for each tracked feature point. The tracker with the maximum number of matches and with a most recent match is chosen as the optimal tracker. The role of the optimal tracker is to assist in updating the tracker of a feature point which had no match. The optimal tracker is also used in estimating the object velocity. To understand the behaviour of the safety system, we used the DoG detector in combination with SURF, BRIEF, and FREAK descriptors, while linBP and iSGM are used as stereo matchers. The novelty in our work is the performance evaluation of a stereo-based collision avoidance system (avoidance by brake warning) in a real collision scenario.

    View record details
  • Incremental Structured ICP Algorithm

    Geng, Haokun; Chien, J; Nicolescu, Radu; Klette, Reinhard (2014-12-08)

    Conference item
    The University of Auckland Library

    Variants of the ICP algorithm are widely used in many vision- based applications, such as visual odometry, structure from motion, 3D reconstruction, or object segmentation. Establishing correct correspon- dences between two sets of 3D points, generated by stereo vision, needs to take those uncertainties into account. We propose a novel variant of the traditional ICP algorithm for solving the mentioned alignment problem. Our method, named incremental structured iterative closest point, aims at improved registrations of 3D points calculated for real world data. Evaluations are carried out by measuring the distances between two in- puts, in both local and global perspectives, and experimental results are visually presented in this paper.

    View record details
  • Comparison of two 3D tracking paradigms for freely flying insects

    Risse, B; Berh, D; Tao, Junli; Jiang, X; Klette, Reinhard; Klaembt, C (2013-10)

    Journal article
    The University of Auckland Library

    In this paper, we discuss and compare state-of-the-art 3D tracking paradigms for flying insects such as Drosophila melanogaster. If two cameras are employed to estimate the trajectories of these identical appearing objects, calculating stereo and temporal correspondences leads to an NP-hard assignment problem. Currently, there are two different types of approaches discussed in the literature: probabilistic approaches and global correspondence selection approaches. Both have advantages and limitations in terms of accuracy and complexity. Here, we present algorithms for both paradigms. The probabilistic approach utilizes the Kalman filter for temporal tracking. The correspondence selection approach calculates the trajectories based on an overall cost function. Limitations of both approaches are addressed by integrating a third camera to verify consistency of the stereo pairings and to reduce the complexity of the global selection. Furthermore, a novel greedy optimization scheme is introduced for the correspondence selection approach. We compare both paradigms based on synthetic data with ground truth availability. Results show that the global selection is more accurate, while the previously proposed tracking-by-matching (probabilistic) approach is causal and feasible for longer tracking periods and very high target densities. We further demonstrate that our extended global selection scheme outperforms current correspondence selection approaches in tracking accuracy and tracking time.

    View record details
  • Egomotion Estimation by Point-Cloud Back-Mapping

    Geng, Haokun; Nicolescu, Radu; Klette, Reinhard (2014-09-15)

    Conference item
    The University of Auckland Library

    We consider egomotion estimation in the context of driver-assistance systems. In order to estimate the actual vehicle movement we only apply stereo cameras (and not any additional sensor). The paper proposes a visual odometry method by back-mapping clouds of reconstructed 3D points. Our method, called stereo-vision point-cloud back mapping method (sPBM), aims at minimizing 3D back-projection errors. We report about extensive experiments for sPBM. At this stage we consider accuracy as being the rst priority; optimizing run-time performance will need to be considered later. Accurately estimated motion among subsequent frames of a recorded video sequence can then be used, for example, for 3D roadside reconstruction.

    View record details
  • Hybrid Filter Blending to Maintain Facial Expressions in Rendered Human Portraits

    Rezaei, Mahdi; Lin, J; Klette, Reinhard (2014)

    Journal article
    The University of Auckland Library

    Artistic rendering of human portraits is different and more challenging than that of landscapes or flowers. Issues are eye, nose, and mouth regions (i.e., facial features) where we need to represent their natural emotions. Shades or darkness around eyes, or shininess at nose tips may negatively impact the rendering result if not properly dealt with. Similarly, a lighter colour around the mouth region caused by lighting might produce some disturbing artefacts. The proposed computerised method attempts to be adaptive to those sensitive areas by utilising a face analysis module. First, the program detects main facial segments and features. Then it utilises a blending of various filtering parameters aiming at an adequate final portrait that represents the subject's original facial expression, while still supporting a non-photorealistic artistic rendering as the perceived impression.

    View record details
  • Look at the Driver, Look at the Road: No Distraction! No Accident!

    Rezaei, Mahdi; Klette, Reinhard (2014)

    Conference item
    The University of Auckland Library

    The paper proposes an advanced driver-assistance system that correlates the driver's head pose to road hazards by analyzing both simultaneously. In particular, we aim at the prevention of rear-end crashes due to driver fatigue or distraction. We contribute by three novel ideas: Asymmetric appearance-modeling, 2D to 3D pose estimation enhanced by the introduced Fermat-point transform, and adaptation of Global Haar (GHaar) classifiers for vehicle detection under challenging lighting conditions. The system defines the driver's direction of attention (in 6 degrees of freedom), yawning and head-nodding detection, as well as vehicle detection, and distance estimation. Having both road and driver's behaviour information, and implementing a fuzzy fusion system, we develop an integrated framework to cover all of the above subjects. We provide real-time performance analysis for real-world driving scenarios.

    View record details
  • Novel Adaptive Eye Detection and Tracking for Challenging Lighting Conditions

    Rezaei, Mahdi; Klette, Reinhard (2013)

    Conference item
    The University of Auckland Library

    The paper develops a novel technique that significantly improves the performance of Haar-like feature-based object detectors in terms of speed, detection rate under difficult lighting conditions, and reduced number of false-positives. The method is implemented and validated for driver monitoring under very dark, very bright, and normal conditions. The framework includes a fast adaptive detector designed to cope with rapid lighting variations, as well as an implementation of a Kalman filter for reducing the search region and indirect support of eye monitoring and tracking. The proposed methodology effectively works under low-light conditions without using infrared illumination or any other extra lighting support. Experimental results, performance evaluation, and comparing a standard Haar-like detector with the proposed adaptive eye detector, show noticeable improvements.

    View record details
  • 3D Cascade of Classifiers for Open and Closed Eye Detection in Driver Distraction Monitoring

    Rezaei, Mahdi; Klette, Reinhard (2011)

    Conference item
    The University of Auckland Library

    Eye status detection and localization is a fundamental step for driver awareness detection. The efficiency of any learning-based object detection method highly depends on the training dataset as well as learning parameters. The research develops optimum values of Haar-training parameters to create a nested cascade of classifiers for real-time eye status detection. The detectors can detect eye-status of open, closed, or diverted not only from frontal faces but also for rotated or tilted head poses. We discuss the unique features of our robust training database that significantly influenced the detection performance. The system has been practically implemented and tested in real-world and real-time processing with satisfactory results on determining driver???s level of vigilance.

    View record details
  • A flexible method for localisation and classification of footprints of small species

    Geng, Haokun; Russell, James; Shin, Bok Suk; Nicolescu, Radu; Klette, Reinhard (2011)

    Conference item
    The University of Auckland Library

    In environmental surveillance, ecology experts use a standard tracking tunnel system to acquire tracks or footprints of small animals, so that they can measure the presence of any selected animals or detect threatened species based on the manual analysis of gathered tracks. Unfortunately, distinguishing morphologically similar species through analysing their footprints is extremely difficult, and even very experienced experts find it hard to provide reliable results on footprint identification. This expensive task also requires a great amount of efforts on observation. In recent years, image processing technology has become a model example for applying computer science technology to many other study areas or industries, in order to improve accuracy, productivity, and reliability. In this paper, we propose a method based on image processing technology, it firstly detects significant interest points from input tracking card images. Secondly, it filters irrelevant interest points in order to extract regions of interest. Thirdly, it gathers useful information of footprint geometric features, such as angles, areas, distance, and so on. These geometric features can be generally found in footprints of small species. Analysing the detected features statistically can certainly provide strong proof of footprint localization and classification results. We also present experimental results on extracted footprints by the proposed method. With appropriate developments or modifications, this method has great potential for applying automated identification to any species.

    View record details