| |
|
|
|
Motion Synthesis on Manifolds in High Dimensions |
|
|
|
A lot of work has been done on how to reuse
motion capture data and minimize the amount of manual work required
in obtaining it. Processing and analyzing the input data, which is
of a high dimensionality, recognizing the data and generating new
motion behavior can be very complicated. Therefore it is necessary
to reduce the data dimensions, while preserving the motion features
embedded in the data. We apply a projection procedure which projects
all points on a lower dimensional manifold, but in higher
dimensional space thus preserving the features among the input
points. This work presents 2 sample applications that can utilize
this method. This was my project for Character Animation class.
This is still an ongoing project.
|
|
|
|
|
|
In this project we were to implement an inverse
kinematics algorithm for controlling a 5 link kinematic chain. The
links were attached at their ends by pin joints and the entire
assemblage moved in a plane. Whenever you click, the arm
automatically deforms to a pose satisfying the constraint. For the
IK algorithm, we were to use the iterative method based on the
pseudo-inverse of the Jacobian. |
|
|
|
Shape to Image Registration |
|
|
|
The goal of this project was to perform a shape
to image registration which is used a lot in medical imaging fields.
We were provided with an initial program and the objective was to
analyze different registrations based on different input parameters
and modify the registration process as needed for a better
registration. More results and images can be seen
here.
|
|
 |
|
|
|
The goal of this project was to analyze the
initial rigid registration of the current algorithm Then better
improvements to the algorithm like better distance functions and/or
better closest point functions were implemented using ITK. More
results and images can be seen
here. |
|
 |
|
|
|
We were provided with a basic implementation of
the eigenfaces algorithm built from ITK modules.
We were to analyze the initial results from different perspectives.
Different constraints on the images were tested. A filter was also
coded in the produce less variability in the images thus
producing less principle components required to reconstruct the
image. More ... |
|
 |
|
|
|
In this project we were to write an OpenGL
program in which the user can explore a virtual world. The user was
to be able to navigate in 3D; translating forwards and backwards,
rotating left and right, and translating up and down. I chose
an under the sea world. I used Bezier Surfaces to draw the fishes
and their fins. I used SkySphere or SeaSphere to get an affect of
the end of the ocean. The floor was infinite meaning that you would
see or feel the whole sea floor tiled, but in implementation it
wasn't. |
|
 |
|
|
|
In this project we were to draw2D fractal plants.
We were to implement and use your own transformation matrices so
that we'll understand how they are used in OpenGL and other graphics
systems. The plant was to define an L-system defined by recursive
functions. The transformations were specified using our own 2D
transformation matrices, and loaded them into the OpenGL MODELVIEW
matrix. We were not to use any OpenGL matrix commands and had to
implement our own matrix stack.
|
|
|
| |
|
|
