Latest Developments and Results in Automatic SCF Counting. Part II: Improved Image Acquisition and Results Obtained

Abstract: The Trashcam project developed a rapid, automated method used to evaluate the seed coat fragment (SCF) potential of fibers from new cotton plant lines. A camera is used to acquire an image of the fiber web and this is then analyzed by image processing. The latest version of the product, has improved image quality by using a scanner for image acquisition. We compare the automated counts obtained with counts made by an 'expert' in 166 cotton web images. Thanks to the new image acquisition technique and an algorithm whose parameters may be adjusted, the results obtained show that the square root of the automated counts are proportional to the square root of the visual counts. The relationship between the two counts obtained in the context of varietal improvement Bars was: square root of automated count = 1.00 square root of the visual count (r=0.965).