Research Rudi Penne

Abstract

The popular pinhole model for imaging sensors and the associated calibration procedures appear to be inadequate for some of the new generation sensor technology. Even for classical RGB cameras, this standard model leads to unstable calibrations, with the need for an extra model to remove lens distortion. We propose line varieties as a unifying modelling for a broad set of sensors. As opposed to other previously published attempts in this direction, we identify the sub-varieties that correspond to real sensors. This enables us to extend interpolation techniques and Gaussian processes, to support sensor calibration from small samples of lines. We aim fundamental contributions to the fields of Line Geometry and Probabilistic Numerics. Our goal is to develop the framework for multi-sensor configurations (laser scanners, IR-cameras,…), providing measurement fusion, using the developed line models, and to achieve accuracy levels for sensor-supported Radiotherapy.

Researcher(s)

• Promoter: Penne Rudi
• Fellow: De Boi Ivan

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Markerless motion tracking became very popular and common since the introduction of the Microsoft Kinect in 2010 in both the gaming community and industry. To use markerless motion tracking in the field of medical rehabilitation, a higher accuracy and reliability is needed. To achieve this goal, we will combine a 2-D skeleton detection algorithm with the data from multiple 3-D cameras. The developed procedure will be validated with the marker-based Vicon system of the M²OCEAN lab and calibrated 3D body scans of subjects in static position. Afterwards, the technique will be implemented on a treadmill to evaluate the gait of a person. To simulate real world gait information, subjects will wear virtual reality glasses. This virtual environment stimulates the brain and influences the gait of a person, which results in extra information compared to stand alone treadmill walking.

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Ribbens Bart
• Co-promoter: Verwulgen Stijn

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In the project we use 3D vision techniques in order to make the assembly process more efficient and safer. This is done by tracking people, products and machines (like cobots) in a manufacturing environment.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Penne Rudi

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

A new generation of digital cameras makes use of emitted light pulses, more precisely the time between the emission and the reception of the reflected pulse, for computing the depth of the viewed object. This "Time-of-Flight" principle is replacing other 3D-scan strategies such as stereovision and structured light. Though the concept and possibilities of a ToF-camera essentially differs from these that are offered by "classical" optical cameras, the computer vision community still falls back on proven methods for calibration and structure-from-motion issues. We propose new techniques, fully exploiting the Time-of-Flight power, avoiding detection and recognition of features in the image. In a further step, we intend to design a new camera model, more general than the familiar pinhole model, providing a uniform framework for both lateral as depth calibration of ToF-cameras. The theory will be validated by simulations and real experiments (executed by a computer driven robot manipulator). Finally, real life applications will be considered, in cooperation with some of our industrial partners.

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Steenackers Gunther
• Fellow: Roios Pedro

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Camera positioning in vision applications is challenging, but of crucial importance. This is definitely the case when the goal is to make a 3D scan. This is because camera positions determine which parts of an object are visible and which measuring accuracy will be achieved. Our ambition is to automatically determine the scan path or positions of the camera during a 3D scan for a known object. To solve this engineering problem we will use mathematical techniques as 'tensor voting' and 'surface fitting'. The final algorithm provides the industry with the following advantages: 1. Faster and more efficient 3D scans 2. More complicated objects can be scanned 3. Automatic scan planning for every type of 3D sensor/camera in one model. 4. Automatic scan planning for specific measurement setups currently used in industry. This reduces the need for expensive experts.

Researcher(s)

• Promoter: Penne Rudi
• Fellow: Bogaerts Boris

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

A new generation of digital cameras makes use of emitted light pulses, more precisely the time between the emission and the reception of the reflected pulse, for computing the depth of the viewed object. This "Time-of-Flight" principle is replacing other 3D-scan strategies such as stereovision and structured light. Though the concept and possibilities of a ToF-camera essentially differs from these that are offered by "classical" optical cameras, the computer vision community still falls back on proven methods for calibration and structure-from-motion issues. We propose new techniques, fully exploiting the Time-of-Flight power, avoiding detection and recognition of features in the image. In a further step, we intend to design a new camera model, more general than the familiar pinhole model, providing a uniform framework for both lateral as depth calibration of ToF-cameras. The theory will be validated by simulations and real experiments (executed by a computer driven robot manipulator). Finally, real life applications will be considered, in cooperation with some of our industrial partners.

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Mertens Luc
• Fellow: Bogaerts Boris

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

This project represents a research agreement between the UA and on the onther hand IWT. UA provides IWT research results mentioned in the title of the project under the conditions as stipulated in this contract.

Researcher(s)

• Promoter: Mertens Luc
• Co-promoter: Penne Rudi
• Co-promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

We focus on 2D point sets with an elliptic shape and 3D point sets with an ellipsoidal shape, e.g. in camera images or a data fusion setting. Noise on these data points forces us to look for robust procedures that derive the quantities we need. Motivating case study: Suppose that the image is taken by a calibrated camera from a ball with known radius, what is the position of this ball relative to the camera?

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Mertens Luc
• Co-promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

This project will focus on the development of a navigation-theory inspired by the navigation behaviour of bats. This theory will be aimed at making possible the construction and use of topologic i.e., 'landmark'-based, environment maps. Both sonar and vision sensors will provide the input for the observation of the environment and the movement through this environment.

Researcher(s)

• Promoter: Peremans Herbert
• Co-promoter: Dhaene Tom
• Co-promoter: Penne Rudi

Research team(s)

• Engineering Management

Project type(s)

• Research Project