Refine
Year of publication
- 1998 (8) (remove)
Document Type
- In Proceedings (3)
- Article (2)
- ZIB-Report (2)
- Book chapter (1)
Language
- English (8)
Is part of the Bibliography
- no (8)
Keywords
- image processing (1)
- segmentation (1)
Institute
Line integral convolution (LIC) has become a well-known and popular method for visualizing vector fields. The method works by convolving a random input texture along the integral curves of the vector field. In order to accelerate image synthesis significantly, an efficient algorithm has been proposed that utilizes pixel coherence in field line direction. This algorithm, called ``fast LIC'', originally was restricted to simple box-type filter kernels. Here we describe a generalization of fast LIC for piecewise polynomial filter kernels. Expanding the filter kernels in terms of truncated power functions allows us to exploit a certain convolution theorem. The convolution integral is expressed as a linear combination of repeated integrals (or repeated sums in the discrete case). Compared to the original algorithm the additional expense for using higher order filter kernels, e.g.\ of B-spline type, is very low. Such filter kernels produce smoother, less noisier results than a box filter. This is evident from visual investigation, as well as from analysis of pixel correlations. Thus, our method represents a useful extension of the fast LIC algorithm for the creation of high-quality LIC images.
Segmentation tools in medical imaging are either based on editing geometric curves or on the assignment of region labels to image voxels. While the first approach is well suited to describe smooth contours at subvoxel accuracy, the second approach is conceptually more simple and guarantees a unique classification of image areas. However, contours extracted from labeled images typically exhibit strong staircase artifacts and are not well suited to represent smooth tissue boundaries. In this paper we describe how this drawback can be circumvented by supplementing region labels with additional weights. We integrated our approach into an interactive segmentation system providing a well-defined set of manual and semi-automatic editing tools. All tools update both region labels as well as the corresponding weights simultaneously, thus allowing one to define segmentation results at high resolution. We applied our techniques to generate 3D polygonal models of anatomical structures.