First Step in Implementing a Feature-based Stereo Algorithm in a Cost Efficient Way using a PMD Sensor

A. Rasool, K. Hartmann, and W. Weihs (Germany)


Stereovision, imaging sensors, modelling, and computer vision


This paper introduces an idea of using a PMD sensor to implement a feature-based stereo algorithms in a cost efficient way. Photonic Mixer Device (PMD) is a semiconductor sensor that has been developed for 3D reconstruction of the scene. A common approach to implement a feature-based stereo algorithm in a cost efficient way is to introduce constraints which reduce the search area. This paper introduces a constraint “initial disparity guess” calculated by using the depth information provided by a PMD. This allows searching a specific feature point of a stereo image in the shorter range of the scan line of corresponding image. The initial results and findings of the proposed system are shown for the 3D reconstruction of a simple object using real data acquired from both PMD sensor and stereo cameras.

Important Links:

Go Back