Space-Efficient Outlines from Image Data via Vertex Minimization and Grid Constraints
✍ Scribed by John D. Hobby
- Book ID
- 102966625
- Publisher
- Elsevier Science
- Year
- 1997
- Tongue
- English
- Weight
- 509 KB
- Volume
- 59
- Category
- Article
- ISSN
- 1077-3169
No coin nor oath required. For personal study only.
✦ Synopsis
the outlines by eliminating noise-induced ''jaggies.'' The When processing shape information derived from a noisy algorithm that generated Fig. 1b tries to do this by emphasource such as a digital scanner, it is often useful to construct sizing smoothness and quality of fit with little regard to polygonal or curved outlines that match the input to within reducing the vertex count [6].
a specified tolerance and maximize some intuitive notions of
The goal of the present work is to retain the advantages smoothness, simplicity, and best fit. The outline description of smooth, well-fitting outlines, but simplify them enough should also be concise enough to be competitive with binary so that they can be stored compactly. What are these adimage compression schemes. Otherwise, there will be a strong vantages? Consider an electronic library project involving temptation to lose the advantages of the outline representation scanned document images as described by O'Gorman [14].
by converting back to a binary image format. This paper pro-When a document is first scanned into the system, the poses a two-stage pipeline that provides separate control over resulting page images need to be processed to remove the twin goals of smoothness and conciseness: the first stage produces a specification for a set of closed curves that minimize noise and correct for the skew angle. The skew comes the number of inflections subject to a specified error bound; about because the lines of text are not likely to be perfectly the second stage produces polygonal outlines that obey the aligned with the scanner's pixel grid. It is important to specifications, have vertices on a given grid, and have nearly correct the skew because small angle rotations lead to the minimum possible number of vertices. Both algorithms are annoying image defects, especially when viewed on a comreasonably fast in practice, and can be implemented largely puter screen. Agazzi et al. [1] have found that a good way with low-precision integer arithmetic.