Subscribe
Issue No.08 - August (1999 vol.21)
pp: 791-797
ABSTRACT
<p><b>Abstract</b>—A breakpoint classification and tuning approach is proposed for the multiprimitive segmentation of planar curves, and cockhead-like graph is suggested to evaluate the multiprimitive segmentation algorithms. The breakpoints are divided into corners and smooth joints and the types of the segments on both sides of a breakpoint are identified. Then, a joint tuning procedure is exercised to merge/split segments and adjust the joint locations. The carefully designed cockhead-like graph includes all possible combinations and parameters of line and arc segments and serves as a benchmark to test the algorithms. The proposed scheme is simple, fast, threshold-free and robust to quantization and preprocessing errors, thus allowing it to be employed in a variety of applications such as matching and recognition. Test against the suggested benchmark and comparison with those in the literature assures the superiority of the method suggested herein.</p>
INDEX TERMS
Multiprimitive segmentation, breakpoint classification, tuning, $k$-curvature, projective height, benchmark.
CITATION
Hsin-Teng Sheu, Wu-Chih Hu, "Multiprimitive Segmentation of Planar Curves-A Two-Level Breakpoint Classification and Tuning Approach", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol.21, no. 8, pp. 791-797, August 1999, doi:10.1109/34.784310