Abstract
In the conventional intersection voting method for 3-D measurement, positions of intersections must be calculated concerning all the combinations of the feature points on image sequence. To reduce the amount of computation required for calculating intersections, this paper proposes a new method which partitions the image into subblocks and detects a motion vector in each subblock. By applying individually the above voting method to each set of subblocks which correspond to one another in the image sequence, computation load can decrease remarkably.
