|Abstrakt:||Area-based stereo algorithms, as the most universal of all stereovision methods, have the potential to become widely used in many industrial sectors, but their relatively low speed halt the usage. This paper describes the method which can significantly increase computing speed. This new method is called false epipolar constraint. It is an addition to epipolar constraint which can reduce the set of possible corresponding points from whole image to a single line. False epipolar constraint in combination with epipolar constraint is able to reduce even this set. This method is most efficient in area-based stereo algorithms but it can be used in any application, where the epipolar constraint is used.