Quantization Error in Stereo Imaging systems

The ability to obtain the accurate three-dimensional position information in the presence of limited sensor resolution is a crucial task in computer vision and other triangulation systems. Sensors for computer processing applications produce sampled quantized data, whose spatial resolution is determined by limits in device technology and bandwidth. In computer vision and photogrammetry, normally stereo camera setups as shown in Figure 1 are used for obtaining 3-D data. In a stereo camera system the two viewing cameras are separated by a distance b along same base line, which is normally taken along the positive x axis. With such a stereo-camera system, two images namely, left image and right image are obtained of any 3-D point. The three-dimensional coordinates of each image point is found by computing the disparities between the corresponding left and right image points. This results in a scanty distribution of reconstructed 3-D points. Designing of any stereo-system is dependent on various parameters like focal lengths of the viewing cameras, the distance of separation between the two viewing cameras, distance of the cameras from the viewing point and interval of image sampling. The relationship