White specks are fiber clusters that appear as small white flecks on dyed fabrics. Their minute size and depth of shade increases the subjectivity and complexity of detecting and/or measuring them. No standard methods exist to quantify the level of white specks. Image analysis was evaluated as a means to measure white specks in order to remove subjectivity from the classification of manual dye defects. This paper provides systematic procedures for the evaluation of white specks on dyed fabrics. Software, hardware, and methods to obtain the white speck count, average size, and percentage white area are discussed in detail. The Optimas 5.2 software program uses a computer, two monitors, a video camera, and a photographic camera stand. The study involved eight plain weave fabrics each with visually distinct levels of white speck content. These fabrics were initially used to identify the system and software best suited for white speck quantification. The manual thresholding method was precise for one operator, but showed significant differences between operators. The fabrics were evaluated for their white speck content using the manual thresholding method (very operator dependent), and again using the new auto-thresholding method to minimize operator input. The auto-thresholding method uses light meters to maintain constant lighting levels and then automatically sets the threshold levels. The system detected different levels of white specks for these fabrics as expected. The auto-thresholding method was more accurate in actual count of white specks than the manual method. This measurement system gives the textile industry a standard tool and procedure to measure fabric appearance. White speck measurements are necessary not to correct the problem on the finished fabric, but as a way to measure changes in levels of white specks as a result of breeding, fields, ginning, and mill processing studies.
