The following seminars are for B.Tech students (work done in spring 2000).
  1. A1: Is Java sound for Machine Vision?. In this seminar, you (taken by Vikram) will study new Java features (part of the Java Media Framework). This builds up on the work done by Nagender on Emerging Java Technologies (see below). You will implement a few basic vision algorithms as a part of the seminar. Background in Java is required. Background in vision is not required.
  2. A2: Image Based Rendering. In this seminar, you (taken by Hemendra) will study how the new paradigm of Image Based Rendering is used in both computer vision and graphics. The work will involve studying a few research papers. Background in graphics useful.
  3. A3: A Maybe topic. Computational Geometry and Computer Graphics. I am reserving this topic to a motivated student who just has to do graphics, but is interested in theory. More than likely, I will NOT offer this topic unless you can convince me otherwise. This seminar is not being taken by any one.
  4. A4: Wavelet Radiosity. (taken by Naga Kiran) In this seminar, you will study about how Wavelets are used to build complex photorealistic rendered images.
The following seminars are for M.Tech students (taken by Nagender, Vinayak Borkar, and Ashwini) and work done in Fall 1999.
  1. S1: Emerging Java Technologies. In this seminar, you will study new Java features (definitely those not in 1.0, very little of 1.1, and some of 1.2 and beyond). The seminar will involve searching for material over the Internet. Note: When asked, you should be able to prove your understanding of the material at the end by running sample demos. (In other words, you cannot just dump available PDF/HTML documents in your seminar). You should also be able to make calculated guesses on why the standards evolved the way they did/do.
    2. Seminar was completed by Nagender
  2. S2: Wavelets for Object Indexing. In this seminar, you will study how wavelets have been used for indexing objects. The seminar will involve studying wavelets in general (through standard text books), and the specific research area of object indexing. In this seminar, tinkering with existing code is optional.
    3. Seminar was completed by Vinayak Borkar
  3. S3: Volumetric Visualization. In this seminar, you will study octree-based methods for volumetric visualization. The seminar will involve researching methods reported in the literature, and the study of three methods that have been discovered locally. This seminar will involve playing around with existing code. This seminar was not taken.
  4. S4: Image Mosaicing. Seminar was completed by Ashwini.
Other seminars done in the past.

Physically-based Modeling: Preeti Gowaikar


For animation in computer graphics, it is desired that the object under consideration looks realistic and natural. The search for realism leads us to the field of Physically-based Modeling.
This report examines physically based models for elastic objects. The underlying principles of physics are those of elasticity theory. A hybrid model for objects that are less elastic is also considered. Turning to rigid bodies, this report looks at a specific problem, viz., collisions: their detection and response.


Radiosity: M.S. Phani


The Radiosity method has emerged as one of the most powerful techniques for realism in computer generated images. Radiosity is most successful in dealing with man-made environments, interiors of offices and the like. Some of the best computer graphics images to date have been produced by the radiosity method.

The generation of an image by the radiosity method consumes large amounts of time and space. The environment or the scene to be rendered consists of surfaces divided into patches and we need to calculate the form factors between these patches. The hemicube method for finding these form factors suffers from effects of aliasing.

In this seminar we examine a new method to find these form factors efficiently within the framework of the hemicube method, but without the effects of aliasing. We further study a rapid hierarchical method for finding these form factors which is based on the concept of substructuring.



Dithering: Anantha Padmanaban K Potty

Nature provides billions of colors, but since memory is expensive, all colors cannot be displayed on a computer display at any given point of time. In this report we investigate a technique called dithering which allows us to create the illusion of a lot more colors than that are actually available.

The report can be obtained by clicking here

Anantha Padmanaban K Potty