Current research projects

25. Open boundary conditions: shallow water waves
    Fredrik Bergholm
    Funding: UU TN-faculty
    Period: 0207-
    Partners: Kristoffer Döös, Dept. of Meteorology, Stockholm University, Stockholm
    Abstract: This is a topic in numerical analysis: Open boundary conditions for shallow water waves. Early reports on the theme from 2000 (MISU, Stockholm) describe simulation experiments using some known techniques, but current research (2003-) focuses on intermittent open boundaries -- a new idea -- presented at a seminar in Sept. 2003, for researchers from TDB and CBA. The goal is to reduce artificial reflections of waves at the open boundary. A preliminary research report on the topic, with 2D experiments, has been written and will be finalized early 2005.

26. New approach to multi- and hyperspectral imaging
    Hamed Hamid Muhammed, Fredrik Bergholm
    Funding: UU TN-faculty, Swedish National Space Board
    Period: 0306-
    Abstract: Despite an impressive evolution of multi- and hyper spectral sensors, there are still very few or crude ways of obtaining an instantaneous multi- or hyper spectral 2D-image of the environment. In particular, in many applications it may be quite useful to have a portable multi spectral camera where instantaneous images are captured in uncontrolled conditions. What would be more convenient than using the digital camera you like, and upgrade it to a multi spectral sensor by adding a color mosaic to the optics? The idea of using color mosaics attached to the CCD is standard, and can of course yield many more spectral bands sacrificing resolution somewhat. However, this approach has several disadvantages. In a digital camera, the manufacturer has already put in a color mosaic, and it is very difficult and expensive to change these color filters. Making color mosaics on the chip requires microscopic filter mosaics, and if the filter mosaic coincides with the glass sealing the CCD chip, the operation must be done in a (camera manufacturer) laboratory.This is slightly inconvenient for people involved in remote sensing, photogrammetry or image analysis, who rather would like to use a non-invasive technique, where technical details on how the CCD chips is built-up should stay secondary concern. How would one avoid "invading the chip" and mounting something permanent or semi-permanent in front of the chip? In this pilot study we have investigated the possibility of placing the filter mosaic (say a plane parallel plate) in some favourable position in the path of light through the lens system. This technique really produces an image that is able to reveal the differences between near-by colors, or between different materials that appear to have similar colors in the RGB-domain; i.e., a sort of a multi spectral image, with a higher spectral resolution than an ordinary RGB-image, is obtained by using this color mosaic. During the year, a patent application on the color mosaic technique has been filed. The patent is financed by UUAB. The robustness of the calculation from image data to multi- or hyper-spectral data (pixel-wise spectra) is analyzed in a report to be submitted 2005 for journal publication. Cooperation with SLU has been established, for tests with field spectra.

27. Efficient algorithms for computer graphics
    Anders Hast, Ewert Bengtsson
    Funding: Dept. of Mathematics, Natural Sciences, and Computing, University College of Gävle; The Knowledge Foundation
    Period: 9911-0404
    Partner: Tony Barrera, Barrera Kristiansen AB, Uppsala
    Abstract: Computer graphics is increasingly being used to create realistic images of 3D objects. Typical applications are in entertainment (animated films, games), commerce (showing 3D images of products on the web which can be manipulated and rotated), industrial design, and medicine. For the images to look realistic high quality shading and surface texture and topology rendering is necessary. Many fundamental algorithms in this field were developed already in the early seventies. The algorithms that produce the best results are computationally quite demanding (e.g., Phong shading) while other produce less satisfactory results (e.g., Gouraud shading). In order to make full 3D animation on standard computers feasible high efficiency is necessary. We are in this project re-examining those algorithms and are finding new mathematical ways of simplifying the expressions and increasing the implementation speeds without sacrificing image quality, see Figure 8. The project is carried out in close collaboration with Tony Barrera at Barrera Kristiansen AB. Several publications was accepted this year and several papers where presented on conferences as follows: One short paper was presented at the WSCG'04 conference in Plzen Czech Rep. And one full paper was published in IEEE Transactions on Visualization and Computer Graphics. Two papers was published at the national conference SIGRAD, which was held in Gävle this year. One of them was a work in progress paper and it was written together with prof. Stefan Seipel and Daniel Wesslen, and this collaboration with the group in Gävle will continue besides the ongoing collaboration with prof. E. Bengtsson and T. Barrera.

    Figure 8: An original Warn spotlight model (left), a new fast soft edge model (middle), and a proposed hard edge model (right)

28. Efficient forest data visualization
    Stefan Seipel
    Funding: County Council in Gävleborg, EU Goal2, North
    Period: 0308-
    Partners: Daniel Wesslén, University College of Gävle
    Abstract: The objective of this project is to develop and assess efficient methods for visualization of information related to forest industry. The goal is to find rendering techniques that are effective in terms of runtime as well as developing graphical representations that can be processed efficiently from a human-cognitive point of view. At the current stage of the project we are investigating techniques for realistic rendering of detail-rich Scandinavian forests. In contrast to previous research in complex plant systems we develop algorithms for procedural modeling, which , based on a parametric description, automatically synthesize typical structures of Scandinavian vegetation. For realistic experience of virtually rendered eco-systems, it is not sufficient to provide high detailed models. In addition to geometric detail, motion of foliage and twigs is an essential visual cue. Traditional techniques for animation of complex plant systems, the model is animated at the host side at given animation frames. Thereafter, updated models must be transferred to the graphics subsystem to be rendered. Current graphics hardware is by all means capable of storing and rendering very high detail plant models. However animation of is hampered due to the limited memory bandwidth between host system and graphics subsystem. Our current research utilizes advanced vertex and pixel shader programming capabilities to perform the entire pipeline of model generation, animation and rendering on the graphics subsystem. Interactive visualizations of realistic Scandinavian forests is an important tool to perform long-term forest planning and in education.

29. Interactive exploration of medical images for visualization on standard PC hardware
    Suthakar Somaskandan, Ingela Nyström, Ewert Bengtsson
    Funding: Sida
    Period: 0409-
    Abstract: Several of the modern imaging systems provide 3D volume information, e.g., CT, MRT, SPECT, and ultrasound. This is very useful since the human body is 3D. Display devices are, however, 2D. Therefore, the 3D image volumes need to be projected onto the 2D computer screens in some way. So far mainly slice by slice projections have been used. There is, however, potential for using much more sophisticated display methods.

    It is particularly useful to device methods where the user interactively can explore the 3D information in the images to reach, e.g., a diagnostic conclusion. Earlier, allowing dynamic interaction with medical image volumes of realistic, clinically useful resolution required very expensive display stations driven by high-performance computers. Today, a PC equipped with a high end standard graphics card can be used quite effectively for that purpose. Such cards typically cost less than 1,000 Euro since they are mass produced for the game markets. But they also offer very interesting potential for use together with appropriate software algorithms for medical visualization. Still much research work is needed to find out the most effective way of using such display facilities for exploring medical image volumes. The research task will be to develop and evaluate such display methods.

30. Co-located information visualization
    Stefan Seipel, Lars W. Pettersson
    Funding: Swedish Defense College, Stockholm
    Period: 0401-0412
    Partners: Swedish Defense College, Stockholm
    Abstract: Advanced graphical visualizations of complex data become increasingly important in decision-making processes where many different stakeholders are involved. Often, these collaborative decision processes can be found in situations where information is gathered and represented in a geographical context. Advanced visualizations which provide 3D visual information to several simultaneous observers call for new display techniques and visualization methods. The objective of this project is to research new methods for visualizing data efficiently for a group of people who interact in the same physical environment. Our research builds upon a novel graphical display system which provides up to eight independent visual channels in the same physical place. With this technology available, new rendering and visualization techniques become feasible that support several people in sharing 3D graphical content. In this project our current research focus is on the design of visualization techniques and consequently on quantitative studies of the humans capacity to utilize 3D graphical content.

31. Colour correction of underwater multi- or hyper spectral data.
    Julia Åhlen, Tommy Lindell, Ewert Bengtsson
    Funding: The Knowledge Foundation
    Period: 0102-
    Abstract: Color restauration of underwater images can be important for many applications, such as marine biology and underwater archaeology.The absorption of light as we descend under water cause production of color corrupted images, see Figure 9 (left). Objects are blue or green colored since shorter wavelengths are absorbed last in the water The aim of the project is to examine the effects introduced by the optical properties of the water column and present a method for removing these negative effects from images. As a first step for this project we estimate a diffuse attenuation coefficient for three wavelengths. The estimation is possible as we are using known reflectance values of a reference gray target that is present on all tested images. To calculate new intensity values we are using Beers Law, where the depth parameter is derived from images that are taken at different depths approximately 50 cm from each other. The suggested method provides a quite effortless and economical way for color reconstruction in degraded underwater images. Another parameter that needs to be taken into account is the image enhancement functions built into the camera. We use a spectrometer and known reflectance standard to obtain the data needed to approximate the joint effect of these functions and the transmittance factor for water mass. This model is used to pre-process the underwater images taken by digital cameras so that the red, green and blue channels show correct values before the images are subjected to correction for the effects of the water column through application of Beer's Law., see Figure 9 (right).

    Figure 9: An underwater color corrupted image (left), and color corrected (right).

    

32. Remote sensing for change detection and monitoring of Case II and lake waters
    Petra Philipson, Tommy Lindell
    Funding: Foundation for Strategic Environmental Research (MISTRA), RESE programme
    Period: 9701-
    Partners: Niklas Strömbeck and Don Pierson, Dept. of Evolutionary Biology, Limnology, UU
    Abstract: The ability to map and monitor water quality parameters in Case II and lake waters is of great environmental interest. Remote sensing is an important constituent for monitoring and an invaluable complement to field observations. Earlier work has been concentrated on finding relations between the water quality variables (Chl, SPIM and CDOM) and remote sensing images from a spectrographic sensor (CASI). However, it is unlikely that the resulting algorithms from these kind of empirical relationships will be sufficiently general to be used in a variety of contexts.
    During this project, a simple bio-optical model was developed and used together with historical water quality measurements from Lake Mälaren, Sweden, to construct algorithms for retrieval of water quality parameters from remote sensing data. These image independent algorithms have been applied to the atmospherically corrected CASI data from Lake Mälaren and Lake Erken and the resulting concentration maps have been validated using ground truth measurements. The results from the validation of the CASI algorithms were satisfying, and the modelled concentrations and absorption coefficients corresponded well to the ground truth measurements, which is very encouraging for the future work. This project was finished with the RESE project ending in December 2003, but got further funding from Swedish National Space Board and has continued within the Department of Limnology by Petra Philipson, now working for Swedepower.

33. New techniques for information extraction from remotely sensed hyper spectral images of Swedish inland waters
    Hamed Hamid Muhammed, Tommy Lindell, Ewert Bengtsson
    Funding: UU TN-faculty, Swedish National Space Board
    Period: 0001-
    Abstract: A substance can be characterised and recognised by its spectral signature. The benefit of hyper spectral imagery is that a sufficient number of narrow spectral bands is available to be able to accurately determine the spectral response at each pixel in the image. A pixel (or a point spectrum) in a hyper spectral image can be considered as a mixture of the reflectance spectra of several substances that can be found in the corresponding imaged region. Independent Component Analysis (ICA) and Principal Component Analysis (PCA) have been used to transform the hyper spectral image data as a first step to get a new set of data that is more suited for further processing than the original data. The next step is to interpret and use the ICA or PCA results efficiently. This can be achieved by using a new technique called Feature-Vector Based Analysis (FVBA) which has been developed earlier in this project.
    The outputs of the transformation step (which are a number of basis vectors and projections of the original data on these vectors) are considered as so called Component-FeatureVector pairs in the subsequent FVBA step. The FVBA task itself is application dependent. But, the common idea of FVBA is to look at the (simpler) Feature Vectors to understand the corresponding (more complicated) Components. FVBA can be used for four main types of applications. In the first two, either well-defined Feature Vectors or well-defined Components are available. The other two types of applications are feature extraction and classification. When studying hyper spectral images of Swedish inland waters, the obtained Feature Vectors and the corresponding Components represent the spectral signatures and the corresponding weight-coefficients images (e.g. the relative concentration maps) of the different constituting substances. The work has resulted in two publications at reviewed conferences, one described FVBA itself, and another one about using FVBA for analysing hyper spectral images. We are now planning to try to apply the methods developed for the hyper spectral crop reflectance data (consisting of 164 spectral bands) in the previously described project, to the much-lower-spectral-resolution hyper spectral images (consisting of 14 spectral bands) of Swedish inland waters.

34. Field measurements, Abisko, Lake Torneträsk
    Tommy Lindell, Fredrik Bergholm
    Funding: UU TN-faculty, Dept. of Infrastructure, KTH, Stockholm
    Period: 0406-
    Partners: Kai Sörensen, Norwegian Institute for Water Research (NIVA); Terry Callaghan, Abisko Scientific Research Station (ANS)
    Abstract: From June 20 to July 2, 2004, spectrometer measurements were performed by Fredrik Bergholm, in birch forest around Abisko Naturvetenskapliga Station (ANS), and spectral mesuremements were performed by Tommy Lindell and Kai Sörensen, in Lake Torneträsk. These field measurements will be part of data base, for the Scandinavian NorSEN project, with start in 2005. The project deals with validation of satellite images, and build-up of ground-based spectral sensors around ANS.

35. HYSENS -- Hyper spectral remote sensing using a new version of ROSIS
    Tommy Lindell, Petra Philipson
    Funding: European Space Agency, Deutsche Zentrum für Luft und Raumfahrt
    Period: 0001-0406
    Partners: Don Pierson, Dept. of Evolutionary Biology, Limnology, UU; Eugenio Zilioli, CNR, Milan, Italy; Province Environment Protection Agency of Trent (APPA), Province Ecological Agency of Verone (ECOV) and Regional Environment Protection Agency, Verone, Italy
    Abstract: ROSIS for Algal Mapping in Lacustrine Environment (ROSALMA). ROSALMA was essentially oriented to a double task:
    • to correlate basic water quality parameters like chlorophyll, suspended sediment concentrations and Secchi disc to the hyper spectral data by using a semi-analytical approach already proved in other geographic conditions and with other hyper spectral devices;
    • to determine the best optical spectral windows for mapping the macrophyte growth, in order to design a possible operational tool to be used for environmental emergencies of this kind, especially in mapping its spatial distribution.
    Lindell & Philipson have participated in the work on Lake Garda, Italy earlier and in the evaluations.
    The focus of the last part of this project was devoted to applications of the MERIS sensor to water quality monitoring together with Kai Sörensen, NIVA, Oslo, and using the experiences gained from the CASI and ROSIS sensors. The Final Report was delivered in June 2004.

36. Digital video and colour camera in remote sensing of water
    Tommy Lindell
    Period: 0001-
    Partners: CNR, Milan, Italy
    Abstract: Test of the usefulness of air-borne digital camera and video for mapping water variables. Lindell has been constructing a holder for the digital video/camera for small aircrafts. Data have been collected from Lakes Erken and Mälaren, and from coral bottoms in Bisceyne National Park. Recently, tests of the usefulness of those images have been performed for the classification of the Swedish coastline.

37. Classification of habitats along the Swedish coast using Landsat images
    Petra Philipson, Tommy Lindell
    Funding: Swedish Environmental Protection Agency
    Period: 0304-0402
    Partners: Cecilia Lindblad and Christina Rappe, Swedish Environmental Protection Agency
    Abstract: The Natura 2000-code is a network of ecosystems and environments for species that the EC-countries have agreed to protect. To be able to chose appropriate areas for protection and for other planning purposes, information about existing habitats and land use is important. For the above stated reasons, as a request from the Swedish Environmental Protection Agency, the entire coastal zone of Sweden has been mapped. Landsat-7 ETM+ images from the Image 2000 database have been used, which resulted in a resolution of 25 meters in the final digital classification. These satellite images are geometrically corrected to RT 90 and can therefore be compared all national Swedish maps. Five classes exist in the classified result: Rock, Stone, Sand, High vegetation and Low vegetation. The result, including report and the digital classification result was distributed to the Swedish Environmental Protection Agency and to all Swedish counties in February 2004.

38. Detecting coral reef bleaching from optical satellites (CORBOS)
    Petra Philipson, Tommy Lindell
    Funding: Foundation for Strategic Environmental Research (MISTRA), RESE programme
    Period: 0001-
    Abstract: Recent dramatic bleaching events on coral reefs have enhanced the need for global environmental monitoring. Remote sensing is an important constituent for monitoring of reefs, and an invaluable complement to field observations. We have been investigating the possibilities and limitations of present high resolution satellites for mapping and monitoring coral reefs, with focus on detection of coral bleaching. The work included investigation of the sensor limitations, the optical properties of the bottom features and understanding of the influence of the atmosphere and water column on the collected remote sensing data. The sensors with the best spatial and radiometric resolution available today, e.g. IKONOS, can be useful for mapping and monitoring of reefs, but they are too costly for global surveys. However, our coral bleaching studies indicate that massive bleaching could be detected even from satellites with lower resolution, like Landsat, SPOT and IRS. They could also be useful for coarser, from a spatial and thematic point of view, global mapping and updating purposes. A field study was performed around Bakers Rendezvous, in Belize in 2002. The project was reported in Ambio in December 2003, but it is continually worked on using local support.

39. The development of a general image analysis software platform
    Bo Nordin, Ewert Bengtsson
    Funding: UU TN-faculty
    Period: 8807-
    Abstract: In recognition of the need in image analysis research to have a good platform for interactive work with digital images, we several years ago started a project with the aim of developing such a platform. The project originally involved some 10 man years of work, which would have been impossible to finance by regular research money. But through a cooperation with a group of companies we co-ordinated our interests of obtaining a good software platform for research with their interest in development of a new software product. Unfortunately, the companies never actively turned the resulting system, which was given the name IMP, into a product. At CBA, however, the IMP system has been used as a software basis for most of the teaching and research in image analysis for the last decade.
    Some years ago, we started a major revision of the system as a ``background task'' for Nordin. The main goal was to re-program the core system in C++ to make it easier to maintain and extend. In 2002, we decided to write a completely new program platform, pixy, based on the new C++ core and with all image analysis functions written in C++ in order to take advantage of the C++-specific language constructs (classes, inheritance, polymorphism, templates, etc.) to enhance the programmer's API and make the code more reusable.
    In pixy, it is easy to add plug-in modules with new functionality and new classes: several such modules have been implemented: MUSE (multivariate segmentation) and filter editors for editing filters in the spatial domain as well as in the Fourierdomain. Pixy was released internally at CBA during 2003 and will be released externally during 2005.