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.
To provide a useful automated tool for measuring overall cotton quality statistics
on which monetary prices are based.
Capabilities:
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.