2011 Annual Report
1a.Objectives (from AD-416)
The objective of this cooperative research project is to redesign an automated fabric inspection system which includes a material handling system for winding rolls of cloth integrated with an imaging system that can quickly, and automatically detect, count and measure dark specks (seed coat fragments specifically) on greige cotton fabrics. The system is expected to replace human evaluation (counting) and to provide reliable and repeatable measurements.
1b.Approach (from AD-416)
The white speck inspection system is already available and needs alteration so it can be used both for white speck and seed-coat inspection. The three major components of the current system are:
Materials Handling device: A fabric inspection machine synchronized with a scanning digital camera, an appropriate lighting source, and a motor driving system to advance and reverse the fabric.
Electrical controls: Consistent lighting which results in an image with high contrast. Camera with scanning technology to acquire the images of the fabrics, grayscale, fabric forward, scanning imaging in sync with the fabric motion as the fabric is advanced.
Software systems: Control the materials handling system and the scanning camera (synchronized), grayscale calibration, analyze the images grabbed from the scanning camera, automatically select the brightness, contrast and minimum pixel size and report speck levels in Excel or in *.txt. Soft ware options to analyze stored images (*.bmp), save the images of fabrics (*.bmp) for visual inspection.
Changes needed for seed coat analysis.
Methods of removing trash particles and leaving seed coat fragments on the fabric:
1) Motorized brushes which brush the loose trash particles from the surface of the fabric, leaving seed coat fragments in the fabric.
2) Air jets which blow loose trash particles from the surface of the fabric, leaving seed coat fragments in the fabric.
3) Both 1 & 2 if needed. Both methods should be tested with different configurations (i.e. angle of fabric as it is cleaned and angle of brushes/jet, brushes and jets separately and together).
4) The brush/jet configuration can be turned on and off so the fabric can be inspected with trash particles and seed coat fragments and later run removing the trash and inspecting for seed coat fragments alone. In the off position, the brushes/jets shouldn’t change the surface of the fabric, so when the system is used for white speck analysis, it can be run without any surface changes.
Lighting should be adjusted or a secondary system installed to best highlight dark specks on the greige fabrics. Consistent lighting which results in an image with high contrast between dark specks and greige fabric.
Software systems: Software that can analyze the images grabbed from the scanning camera using a dark-speck detection algorithm for reliable, repeatable measurements of dark specks on greige cotton fabrics (number, size and percent dark coverage of seed coat fragments and trash). Automatically select the brightness, contrast and minimum pixel size removing subjectivity in the measurements and speeding up the process and report results in Excel. Soft ware options to analyze stored images (*.bmp), save the images of fabrics (*.bmp) for visual inspection. Report if the test was run with Trash removal system on or off.
The Autorate system, which was modified for dark specks caused by Seed Coat Fragments (SCF) from white specks (a dye defect), was used to measure SCFs on fabrics. The results from the Autorate were related to hand counted SCFs on the fabric and the Seed Coat Neps (SCN) and Neps measured by AFISPro. Woven fabrics from the US Extreme Variety Study (US EVS) were evaluated by hand counting the dark specks and by image analysis using the modified Autorate system. Strong correlations were found between the hand counted dark specks and the Autorate counted dark specks, which indicates the Autorate to be an accurate measure of dark specks on fabric. These new fabrics had some smaller SCF than previous studies and the minimum pixel size had to be changed for the system to be accurate. New versions of the Autorate for dark specks (SCFs) were evaluated. However, Version 5 (5 pixels is the minimum number of pixels used to indicate dark specks) was the best of the newer versions. We also used the US EVS samples to determine minimum sample size and minimum number of camera flashes.
The methods used to monitor activities for this agreement were annual reports, technical visits/e-mails/interactions, presentations at scientific and industry meetings, and publications.