In a manufacturing environment, objects are often presented to inspection systems via conveyor belts. If multiple snapshots of a moving object are taken by a fixed camera, the motion of the belt provides the necessary stereo disparity. Moreover, it also guarantees that the disparity occurs only along epipolar lines. This method is called motion stereo.A simple algorithm for calculating depth from motion stereo was initially implemented which makes the assumption that the incremental disparity is less than the minimum distance between edges. To relax this constraint, a more general multi-scale algorithm is developed. Reduced images are matched first and this guides the matching of subsequent images. The parallel and hierarchical algorithm facilitates a tight pipeline and interactions among multiple motion stereo images and their multi-scale versions in a combined bottom-up and top-down manner in a pyramidal framework.The two algorithms have been tested on the SFU pyramidal vision machine which was recently completed for real-time computer vision applications. The total processing time was 50 ms per image for the simple algorithm and approximately 0.5s per image for the multi-scale algorithm.