Calibration of Consumer Depth Sensors
Low cost consumer depth sensors like Kinect and Kinect2 are popular sensors to obtain RGB-D image data. This data consists of a video stream of images and a corresponding video stream of depth data. The data can be used for example as an input for SLAM systems that compute a mapping and reconstruction of complex scenes. To obtain accurate representations of the scene additional calibration of systematic errors of these consumer sensors is necessary.
One extremely popular RGB-D sensor choice has been the Kinect1 sensor. The depth data obtained with this sensor is superimposed with multiple geometric errors and many calibration methods to compensate for these errors have been proposed. Due to the complexity of the geometric errors of the Kinect1 sensor, results are usually still not satisfactory. The Kinect2 sensor uses a completely different method to obtain depth information and possible geometric corrections are still not well known.
This thesis should investigate the calibration and the characteristics of the geometric errors of the Kinect2 sensor.
The thesis will be conducted in the scope of the COMET-project Vision+ - Integrating visual information with independent knowledge (http://www.comet-visionplus.at).
C/C++, computer vision and math skills
this mastertheses at JOANNEUM RESEARCH DIGITAL is scheduled for approx. 6 month.
JOANNEUM RESEARCH offers interested students cooperation with all Austria but also European or non-European educational institutions and is willing to assist you looking for a supervisor.