Andreas Östlund

Master Thesis Project

andreasostlund — Thu, 07 Feb 2008 20:16:42 +0000

As a part of my master thesis project me and my colleague have created an application that visualizes and classifies motion capture data from an ergonomic perspective. This is a screenshot from it:

More info about this application and the rest of the thesis work will follow…

Procedural Image Project

andreasostlund — Thu, 07 Feb 2008 20:06:12 +0000

This is a project I did for the course Procedural Images. It is written i C++ and uses the wxWidgets toolkit and OpenGL and it is a 3D procedural terrain renderer. I use different noise funtions to generate the terrain and the clouds. GLSL was used to create the colored ground, which is based on altitude and slider settings. The shape of the terrain can be altered in real time with different sliders.

Infoviz project – elViz

andreasostlund — Thu, 07 Feb 2008 19:57:25 +0000

This is a screenshot from my infoviz application created for the course Information Visualization. The application is created using C# and the GAV framework and visualizes Swedish energy consumtion. It also includes an implementation of K-means clustering.

Visual Data Mining Project

andreasostlund — Thu, 07 Feb 2008 19:22:39 +0000

A project done in my fourth year together with another student. In this project we compare and ultimately combine two straight forward implementations of outlier detection. Visual and Distance based. Our main findings were:

Visual Outlier Detection can if correctly implemented be an effective method forfinding outliers. However with large datasets of highly multivariate data, problems can arise due to the simple fact that there is too much information in the same place in on the computer screen. This combined with the fact that it is difficult to map many tuple attributes at the same time , can cause major difficulties in finding outliers. Techniques studied are 2D and 3D Scatter Plot as well as the Parallel Coordinates Plot (PC).

Distance Based Outlier Detection. From our implementation we found that distance based outlier detection works quite well and can identify both natural and unnatural (faulty data) outliers but the fact that it is non graphical can make the results difficult to understand.

When Combining Visual and Distance Based Outlier Detection we found that the benefits of both methods complement each other very nicely. Giving an robust and intuitive method of finding outliers in multi variate data.

Simulation Project

andreasostlund — Wed, 25 Apr 2007 20:10:55 +0000

A project done with 3 other students in my third year. Euler’s time integration, OpenGl and C++ was used to simulate a physically correct inclined snowboard turn. It features a realtime gametype control of the snowboard and bullet time

AI Project – SMS Dictionary

andreasostlund — Wed, 25 Apr 2007 20:09:27 +0000

This application was created as a part of an AI course. It simulates a smart text message dictionary used in mobile phones and was written in Lisp.

Modelling and Animation Project

andreasostlund — Wed, 25 Apr 2007 20:08:07 +0000

A project done with 4 other students in my second year. The entire 3D movie, featuring an explosive car chase, was created with 3D-Studio Max.

Volume Rendering of a Head

andreasostlund — Thu, 15 Mar 2007 17:08:19 +0000

I just completed the final lab assignment in the Scientific Visualization course. It was created using Visualization Toolkit (VTK) and features volume rendering of a CT scanned head.

(texture based) (raycast) (simple iso surface)

Information Visualization

andreasostlund — Mon, 19 Feb 2007 21:41:27 +0000

I just finished the third lab in the infovis course. It is a visualization of swedish municipality data. It is visualized using glyphs , color mapping and parallell coordinates. The program is written in C# and uses the GAV framework, which is being developed at NVIS here in Norrköping.

Busy with School

andreasostlund — Sat, 10 Feb 2007 15:18:53 +0000

So the winter break is over and school is hectic as usual. I’m taking two courses now, Image based rendering and Infomation Visualization.

For the second lab in IBR we implemented (in Matlab) a simple Image based lightning model and did a lookup to produced a car reflecting the background lightning:

We also created a panoramoc longitude latitude map from a sphere (see right)