Fabric wrinkling and pilling evaluation by stereovision and three-dimensional surface characterization
Wrinkling and pilling caused in wear and care procedures are vital performance characteristics of fabric. The advance of three-dimensional (3D) imaging techniques has made it possible to develop a convenient, reliable and low cost tool for automatic and efficient evaluation of fabric wrinkling and pilling. We suggest that 3D imaging and measurement system can provide a convenient, accommodating and comprehensive mean to fabric surface assessment. A 3D imaging system based on stereo vision technology is developed. To make it more affordable and portable, the system consists of a pair of consumer grade high resolution digital cameras with mounting hardware. The system is calibrated with classic camera calibration technique. The calibration procedure is relatively complicated, but there is no need to repeat frequently as long as the relative positions between cameras are not changed. In this system, image acquisition can be completed in less than one second. This efficient surface capturing feature is important for a large amount of measurement tasks. However, the computation in stereo vision is complex and intensive, thus it remains a challenge. A two-phase multi-resolution stereo matching algorithm is developed. In the first phase, a discrete disparity map is generated by block matching. In the second phase, local least-squares matching is performed in combination with global optimization within a regularization framework, so as to ensure both accuracy and reliability. To make the 3D imaging system ready for practical use, detection and measurement modules for wrinkling and pilling were developed to take advantage of the depth information in the 3D surface data. The practical feasibility of the 3D imaging system in fabric surface assessment was demonstrated in comparison with human visual ratings. The results showed agreement between the 3D automatic assessment and subjective visual assessment.