Refereed conference proceedings

1. Estimating Fibre Twist and Aspect Ratios in 3D Voxel Volumes
   Author: Aronsson, M.
   Conference: International Conference on Pattern Recognition (ICPR 2002)
   Editors: Kasturi, R.; Laurendeau, D.; Suen, C. (ed.)
   Publisher: IEEE Computer Society, vol. I, 218-221, 2002
   Abstract: Aspect ratios and twist measures can help us characterise paper fibre properties. Such measures, that need volume data, are presented here. To test the developed methods, simulated voxel volumes of fibres were created by defining each fibre as a spline curve with an elliptical cross sectional shape and a constant twist per length unit. This allows us to directly compare the measurements from the voxel volume to the properties of the original spline fibres.

2. Ring Shaped Object Detector for Non-Isotropic 2D Images Using Optimized Distance Transform Weights
   Authors: M. Aronsson, I.-M. Sintorn
   Conference: International Conference on Image Processing (ICIP 2002), Rochester, NY, USA
   Publisher: IEEE Publications, pp. 985-988, 2002
   Abstract: A detector for finding ring shaped objects occurring in clus-ters in 2D images with non-isotropic pixel dimensions have been developed. The rings are characterized as having a closed border and a void interior. We assume that the thick-ness of the rings should be approximately constant and their void shape smooth. We initially find candidates that could be possible ring voids, then by using a distance transform (DT) based technique, a DT histogram of the surrounding material is created. If the DT histogram fulfills certain con-straints, a ring has been found. A second DT based step will mark the pixels surrounding the void with a unique label. By this approach we impose only loose constraints on the shape of the rings. In addition to the ring detector itself, the paper illustrates that if proper DTs are used, non-isotropic images can be analysed without interpolation to a square grid.

3. Segmentation of Liver Images by Texture and Genetic Snakes
   Authors: Ballerini, L.; Bocchi, L. (1)
   (1) Dept. of Electronics and Telecommunications, University of Firenze, Italy
   Conference: European and Biological Engineering Conference (EMBEC'02), Vienna, Austria
   Proc. pp. 1096-1097, 2002
   Abstract: The liver is a common site of metastatic disease. The diagnosis of liver metastases in patients at high risk is straightforward, but estimating the gravity is more problematic. We investigate the use of CT scans. The first problem is to segmenting the liver itself in the CT scan, the second is segmenting the metastases in the liver to evaluate their relative volume. In this pre-study, we investigated a new method to segment the liver based on textural properties of the image and active contour models.

4. Determination of Pores in Pig Meat Images
   Authors: Ballerini, L.; Bocchi, L. (1); Hullberg, A. (2)
   (1) Dept. of Electronics and Telecommunications, University of Firenze, Italy
   (2) Dept. of Food Science, SLU, Uppsala
   Conference: International Conference on Computer Vision and Graphics, Zakopane, Poland
   Proc. vol. 1, pp. 70-78, 2002
   Abstract: In this paper we present an image processing application for quantification of pores in pig meat images. We used digital camera photographs of slices of tumbled and non-tumbled processed loins of different RN genotype. We describe a method to extract and count the number and sizes of pores. The aim is to study the relationship between the amount of pores and process condition and genotype. Results show that image analysis can be useful in this field.

5. Fuzzy Border Distance Transforms and their Use in 2D Skeletonization
   Authors: Borgefors, G.; Svensson, S.
   Conference: International Conference on Pattern Recognition (ICPR 2002)
   Editors: Kasturi, R.; Laurendeau, D.; Suen, C.
   Publisher: IEEE Computer Society, vol. I, 180-183, 2002
   Abstract: Segmentation is always a difficult task in image analysis. In this paper, we propose a solution to computing distance transforms in images with fuzzy object borders. The difference from a standard distance transform is in the initialisation, which takes the fuzziness of the border into account. As an example of its usefulness, the new fuzzy border distance transform is used in skeletonization.

6. Surface Area Estimation of Digitized 3D Objects using Local Computations
   Authors: Lindblad, J.; Nyström, I.
   Conference: Discrete Geometry for Computer Imagery (DGCI 2002)
   Editors: Braquelaire, A.; Lachaud, J.-O.; Vialard, A.
   Publisher: Springer-Verlag, Lecture Notes in Computer Science 2301, pp. 267-278, 2002
   Abstract: We describe surface area measurements based on local estimates of isosurfaces originating from a marching cubes representation. We show how improved precision and accuracy are obtained by optimizing the area contribution for one of the cases in this representation. The computations are performed on large sets (approximately 200,000 3D objects) of computer generated spheres, cubes, and cylinders. The synthetic objects are generated over a continuous range of sizes with randomized alignment in the digitization grid. Sphericity, a scale invariant measure of compactness, allows us, in combination with the improved surface area estimate, to distinguish among the test sets.

7. Area of and Volume Enclosed by Digital and Triangulated Surfaces
   Authors: Nyström, I.; Udupa, J.K. (1); Grevera, G.J. (1); Hirsch, B.E. (1)
   (1) MIPG, Dept. of Radiology, University of Pennsylvania, Philadelphia, USA
   Conference: SPIE Medical Imaging 2002: Visualization, Image-Guided Procedures, and Display
   Editor: Mun, S.K.
   Publisher: SPIE Proceedings 4681, pp. 669-680, 2002
   Abstract: We demonstrate that the volume enclosed by triangulated surfaces can be computed efficiently in the same elegant way the volume enclosed by digital surfaces is computed by digital surface integration. Although digital surfaces are good for visualization and volume measurement, their drawback is that surface area measurements are inaccurate. On the other hand, triangulated surfaces give more accurate surface area measurements, but volume measurements and visualization are less efficient. The T-shell data structure previously proposed retains advantages and overcomes difficulties of both the digital and the triangulated approaches. We create a lookup table with area and volume contributions for each of the 256 Marching Cubes configurations. When scanning the shell (e.g., while creating it), the surface area and volume are incrementally computed by using the lookup table and the current x co-ordinate, where the sign of the x component of the triangle normal indicates the sign of the volume contribution. We have computed surface area and volume for digital and triangulated surfaces for digitized mathematical phantoms, physical phantoms, and real objects. The computations show that triangulated surface area is more accurate, triangulated volume follows digital volume closely, and that the values get closer to the true value with decreasing voxel size.

8. Weighted distance transforms for images using elongated voxel grids
   Authors: Sintorn, I.-M.; Borgefors, G.
   Conference: Discrete Geometry for Computer Imagery (DGCI 2002)
   Editors: Braquelaire, A.; Lachaud, J.-O.; Vialard, A.
   Publisher: Springer-Verlag, Lecture Notes in Computer Science 2301, pp. 244-254, 2002
   Abstract: In this paper we investigate weighted distance transforms in 3D images using elongated voxel grids. We use a local neighbourhood of size 3×3×3 and assume a voxel grid with equal resolution along two axes and lower along the third. The weights (local distances) in the local neighbourhood are optimized by minimizing the maximum error over linear trajectories, which is a completely digital approach. General solutions are presented, as well as numerical solutions for the cases when the voxels are 1.5 and 2.58 times longer in one direction. Integer solutions for both real and integer scale factors are presented. As an application example, the medial axis of an object is computed in an image with elongated voxels and compared to the medial axis computed on the same image interpolated to equal resolution along all axes.

9. Using Grey-Level and Distance Information for Medial Surface Representation of Volume Images
   Authors: Svensson, S.; Nyström, I.; Sanniti di Baja, G. (1); Arcelli, C. (1)
   (1) Istituto di Cibernetica, National Research Council of Italy (CNR), Arco Felice (Napoli), Italy
   Conference: International Conference on Pattern Recognition (ICPR 2002)
   Editors: Kasturi, R.; Laurendeau, D.; Suen, C.
   Publisher: IEEE Computer Society, vol. II, pp. 324-327, 2002
   Abstract: A medial surface representation of a grey-level volume image is computed. The foreground is reduced to a subset topologically equivalent to the initial foreground and mainly consisting of surfaces centred within regions having locally higher intensities, here, regarded as more informative. This result is obtained by combining distance information with grey-level information. A surface skeleton is first computed, where excessive shortening is prevented by a regularity condition defined on the distance transform. The structure of the surface skeleton is then simplified by removing some peripheral surfaces, so obtaining the desired medial surface representation.

10. Acquisition and Colour Correction of Underwater Multispectral and Hyperspectral Images
    Author: Åhlén, J.
    Conference: 7th International Conference of Remote Sensing for Marine and Coastal Environments 2002, Miami, Florida, USA
    Publisher: Veridian ERIM International, CD-rom, 2002
    Abstract: The examination of image processing techniques for dealing with image enhancement in underwater conditions is important for scientists involved with marine environments. One application could be a study of archaeological sites in various oceans of the world. Another application is a different approach to study problems observed on corals such as bleaching. Prominent blue colour of clear ocean water, apart from sky reflection, is due to selective absorption by water molecules. Due to this nature of underwater optics, red light diminishes when the depth increases, thus producing blue to grey like images. No studies that could help efficiently reduce the negative effects of scatter and light absorption has been done on multi spectral or hyper spectral data taken under the water. Investigation of multi or hyper spectral data will give us more colour channels to work with. An image-processing algorithm that takes in pre-processed colour channels and enhance some characteristics depending on application is the result we would want to achieve. I will present a technique for acquisition of hyper and multivariate data under the surface in saltwater. The colour constancy problem will be discussed considering particular application.