A protocol for performance evaluation of line detection algorithms
Machine Vision and Applications - Special issue on performance evaluation
Empirical Performance Evaluation of Graphics Recognition Systems
IEEE Transactions on Pattern Analysis and Machine Intelligence
Performance Evaluation of Line Drawing Recognition Systems
ICPR '00 Proceedings of the International Conference on Pattern Recognition - Volume 4
The third report of the arc segmentation contest
GREC'05 Proceedings of the 6th international conference on Graphics Recognition: ten Years Review and Future Perspectives
Graphics recognition: the last ten years and the next ten years
GREC'05 Proceedings of the 6th international conference on Graphics Recognition: ten Years Review and Future Perspectives
Salt and Pepper Noise Removal from Document Images
IVIC '09 Proceedings of the 1st International Visual Informatics Conference on Visual Informatics: Bridging Research and Practice
| Hi-index | 0.00 |
Correct detection of line attributes by line detection algorithms is important and leads to good quality vectors. Line attributes includes: end points, width, line style, line shape, and center (for arcs). In this paper we study different factors that affect detected vector attributes. Noise level, cleaning method, and vectorization software are three factors that may influence the resulting vector data attributes. Real scanned images from GREC'03 and GREC'07 contests are used in the experiment. Three different levels of salt-and-pepper noise (5%, 10%, and 15%) are used. Noisy images are cleaned by six cleaning algorithms and then three different commercial raster to vector software are used to vectorize the cleaned images. Vector Recovery Index (VRI) is the performance evaluation criteria used in this study to judge the quality of the resulting vectors compared to their ground truth data. Statistical analysis on the VRI values shows that vectorization software has the biggest influence on the quality of the resulting vectors.