Two-dimensional contour matching algorithm based on multiscale space

摘要

The contour matching is a hot issue and a difficult problem of image processing, and the accuracy and the efficient of the algorithms are the most two critical factors. To obtain the most likely matching part,a curve matching algorithm that based on the multiscale space is proposed. Firstly the line length and the angle are calculated as the feature vectors, secondly by setting different angle threshold to reduce the number of feature points, lastly the contour matching is measured by hausdorff distance. Experiments show that the algorithm is effective and stable.