The marker-based target can also be used to reduce the amount of input data for segmentation. Besides, feedback about the image geometry is possible before further computations take place. This may be helpful, because rectification of the entire image is not efficient. For image regions, rectification is feasible, however, because the amount of input data is much smaller. In the problem at hand, a homography between points on the image plane and points lying on a plane defined by the dimensions of the target can be computed. This bijective mapping of coordinates between planes can be represented by an on singular matrix of size 3x3 with 8degrees of freedom (see e.g. the work of Hartley and Zisserman [6]). For the computation (estimation) of the 3x3 Matrix HIW (HWI =HIW−1)at least n=4 point correspondences must be found between image plane and the corresponding world plane. In consideration of a possibility to realize such a rectification procedure efficiently, the minimum number of 4 points is chosen. They correspond to the corners of the reference rectangle of the marker-based target. Thus, a rectangular region of interest (ROI) can be defined in the image plane using the known location and dimensions of the checkerboard area and the previously computed homography.