Digital Signal Processing and Pattern Recognition
One of my major strengths is my experience in digital signal processing and image and signal pattern recognition, and my work & academic coursework focused on these areas.
While at GE I did research on model-based image pattern recognition methods for aircraft identification, and my Masters Thesis was on the AI techniques applied to one-dimensional signal spectral estimation.
Since then I've worked on the following pattern recognition and digital signal processing projects:
In the process I've gained a lot of experience with MATLAB and the Signal Processing and Fixed-Point Filter Design Toolboxes and TI's Code Composer Studio.
- MATLAB Pattern Recognition Toolbox for one-dimensional signal classification
- Developed a classifier for diabetic laser doppler blood perfusion waveforms
- Designed & analyzed fixed-point digital filters for microcontrollers
- Demonstrated a DSP-based laser doppler perfusion waveform processing system based on the TI 6713 DSK.
Software Expertise Associated with Signal Processing:
I've coded my projects in MATLAB and C under the Unix and Windows operating systems, and maintained that code using SCCS and Perforce.
Other software accomplishments include:
- Developing a windowed user interfaces under UNIX (SunView/X-Windows) for analyzing large images - used internally within the GE Military & Data Systems Operation.
- Mapping image pattern recognition algorithms onto SIMD and MIMD parallel processing systems.
- Writing a signal analysis user interface with the MATLAB GUI tools.
Technical Details:
The page banner above is an eight-second spectrogram of a Northern Mockingbird. The recording came from a USGS bird song site. I converted the .MP3 file to a .WAV file using Sound Forge Audio Studio and then used this MATLAB M-file to calculate the spectrogram and compare it with the time series.