Range Imaging: Investigation, Calibration and Development
Published in 2008
Book
Paperback/Softback
142 pages
978-3-906467-72-6 (ISBN)
Unfortunately, price unknown
Article not available
Description
In recent years, numerous sensor systems for the capturing of three-dimensional environments and objects have become available. Besides laser scanners and geodetic total stations, stereo vision and triangulation-based systems have to be exemplarily named here. Especially laser scanners have become state-of-the-art regarding speed and accuracy with respect to their ability to acquire objects up to a size of several tens of meters. A main drawback of laser scanners is their sequential mode of operation. They measure point by point. A few years ago, a new technology was developed to full functionality which is able to capture the environment simultaneously with a high resolution. So-called range imaging (RIM) or flash ladar cameras, which are based on digital imaging technology, merged with the ability to measure the distance to the corresponding object point in each pixel. Distance measurement is either based on the direct or indirect Time-of-Flight principle. Due to its parallel acquisition with up to video frame rate, RIM cameras are even able to capture moving objects. With respect to the optical dependencies, 3-D coordinates of the captured scene are derived. The nominal precision of the distance measurement is a few millimeters. RIM could become the technology of choice for many applications if the properties and characteristics become stable and predictable. Automotive, robotics, and safety systems can be named, for example. Significant deviations between nominal and measured coordinates occur in a range of several centimeters. Only intensive investigations can help to reach the theoretical limitations here.
This thesis deals with several aspects which affect the measurements of RIM cameras. First, a short introduction into the basic technologies that are associated with RIM is presented. Besides imaging and distance measurement methods, two basic principles of RIM are distinguished. Furthermore, the focus is laid on the specific limitations. During this work three different cameras have become available: the SwissRanger SR-2 and the SR-3000 from CSEM / MESA Imaging (Switzerland) and later on the 3k-S from PMDtec (Germany). These three cameras are based on the indirect Time-of-Flight principle and are equipped with different sophisticated features. Besides integrated calibration and correction functionality, the suppression of background illumination is one of the main features. However, these cameras are only intended to be highly developed demonstrators. An adaption to the specific application areas, like automotive or robotics, leads to specialized properties according to the desired claims.
The analysis of the existing camera types helps to understand the technology more closely. The raw data of the analyzed cameras is not more accurate than a few centimeters. In order to investigate the properties of the available cameras, special experimental setups had to be developed. The main part of this work deals with the investigation and calibration of the components of RIM cameras. The geometrical deviations of the optical system are addressed by means of a photogrammetric camera calibration. The distance measurement system is analyzed with respect to the deviations and statistics. Thus, limitations of both precision and accuracy are indicated. Besides the influences of the scattering effect, integration time, emitting system, and angle of incidence, target reflectivity, external and internal temperature, and finally linearity and fixed-pattern noise are discussed. Further on, an approach for a system calibration process is presented. Due to the complexity of the influencing parameters, a complete correction of the measurement data with respect to the diverse influencing parameters has not been reached. But the highly systematic dependencies promise sophisticated calibration routines in the future. This work contributes to the understanding of the sensors.
Nevertheless, the investigated influences of temperature on the distance measurement accuracy, which is indicated as a measure for the deviation between true and nominal value, have been significantly reduced by an uncoupling of the distance measurement and the external and internal temperature by means of a relative measurement setup. The introduction of an internal reference light path helps to reduce the temperature's influence on the distance data to a large degree. The experimental setup and the proof of the functionality complete this work.
The results of the numerous investigations will help to increase the accuracy of RIM cameras, especially vital for several applications, in need of improved accuracies. It has been shown that the theoretical limits lie within reach with help of suitably sophisticated calibration procedures.
This thesis deals with several aspects which affect the measurements of RIM cameras. First, a short introduction into the basic technologies that are associated with RIM is presented. Besides imaging and distance measurement methods, two basic principles of RIM are distinguished. Furthermore, the focus is laid on the specific limitations. During this work three different cameras have become available: the SwissRanger SR-2 and the SR-3000 from CSEM / MESA Imaging (Switzerland) and later on the 3k-S from PMDtec (Germany). These three cameras are based on the indirect Time-of-Flight principle and are equipped with different sophisticated features. Besides integrated calibration and correction functionality, the suppression of background illumination is one of the main features. However, these cameras are only intended to be highly developed demonstrators. An adaption to the specific application areas, like automotive or robotics, leads to specialized properties according to the desired claims.
The analysis of the existing camera types helps to understand the technology more closely. The raw data of the analyzed cameras is not more accurate than a few centimeters. In order to investigate the properties of the available cameras, special experimental setups had to be developed. The main part of this work deals with the investigation and calibration of the components of RIM cameras. The geometrical deviations of the optical system are addressed by means of a photogrammetric camera calibration. The distance measurement system is analyzed with respect to the deviations and statistics. Thus, limitations of both precision and accuracy are indicated. Besides the influences of the scattering effect, integration time, emitting system, and angle of incidence, target reflectivity, external and internal temperature, and finally linearity and fixed-pattern noise are discussed. Further on, an approach for a system calibration process is presented. Due to the complexity of the influencing parameters, a complete correction of the measurement data with respect to the diverse influencing parameters has not been reached. But the highly systematic dependencies promise sophisticated calibration routines in the future. This work contributes to the understanding of the sensors.
Nevertheless, the investigated influences of temperature on the distance measurement accuracy, which is indicated as a measure for the deviation between true and nominal value, have been significantly reduced by an uncoupling of the distance measurement and the external and internal temperature by means of a relative measurement setup. The introduction of an internal reference light path helps to reduce the temperature's influence on the distance data to a large degree. The experimental setup and the proof of the functionality complete this work.
The results of the numerous investigations will help to increase the accuracy of RIM cameras, especially vital for several applications, in need of improved accuracies. It has been shown that the theoretical limits lie within reach with help of suitably sophisticated calibration procedures.