1. To identify trash particles within a cotton sample and make a determination about which ginning steps will produce the cleanest cotton while minimizing damage to the cotton fibers.
2. To provide a useful automated tool for measuring overall cotton quality statistics on which monetary prices are based.

• Training mode with user interaction.
• Operation/trash identification mode.
• Backward Propagation Neural Network, (distinguishes between bark/grass, stick, leaf, pepper trash, and user defined trash types)
• Pixel Calibration --particle data can be reported in real-world measurements or pixels.
• Automatic camera integration and lighting level unit adjustment using color tiles for reference --for accurate color measurement.
• Create sophisticated customizable reports using statistics such as percent trash and the amount of each type of trash found.
• Accurate color mode determines the lint color values in the RdAB color space, for cotton quality control.
• Per image scaling automatically compensates for camera and light level drift over long periods of time to ensure accurate color measurements.
• Continuously updated system and calibration summaries
• Four possible image sources: TWAIN Device, Sony DXC390 Camera, Duncantech/Redlake MS3100, or image files from disk.
• Several optional utilities to control the camera, adjust the lighting unit, and obtain scientific measurements to measure system accuracy.
• LUT mode to ensure uniform lighting distribution.
• Flat Field Correction compensates for non-uniform lighting conditions.
• Voice synthesis provides audio feedback about the current sample and repetition during tests.
• A multiple user system simplifies the option screens and processing procedures for less advanced users, and adds more detailed control for administrators.

Screen Shots:

Trash Identification	Neural Network Training	Auto-Lighting Calibration	Pixel Calibration
Ginning Facility	Utility Windows	Offline Testing Hardware	Offline Testing Hardware
Splash Screen	Real-time Statistical Summaries	LUT Function	Option Screens

Research Pulications (Dr. Murali Siddaiah and Michael Lieberman)