Fuzzy Doc Recognition With Color Space Segmentation and Matching

A small, fuzzy, and stuffed ``Doc'' from Snow White and the Seven Dwarves is to be recognized by the machine vision system described in this paper. Problem constraints are purposefully vague; Doc is to be recognized in a veriety of poses in clutter. A method of color segmentation in the hue-saturation-value (HSV) color space is employed to segment parts of Doc from the clutter, which are bound with a minimum enclosing box of the contours extracted from these segmented blobs. A rudimentary matching algorithm is described.

Implementation and Comparison of Radix-2 and Radix-4 FFT Algorithms

Bullshit paper I wrote for the bullshit DSP class I took. The FFT code is good, though. I went the recursive Cooley-Tukey route, and I think it turned out just peachy. There's a version included at the end of the paper and a newer one here: fft.c which precalculates the twiddle factors for a much needed speedup! Check it out!

Moment Representation of Blobs in 2-D Intensity Images

Short paper on an early fun form of machine vision using moment invariants. It may suck, but you've gotta love those cutlery images.

My Resume

You need me. You may not even realize it yet. But, believe me. You do.

Master's Thesis: An Information Theoretic Hierarchical Classifier for Machine Vision

This is it: the big mama. Binary decision trees + information theory + pattern recognition + machine vision = done with school.