A Practical Method for Detecting Fluff Quality of Fabric Surface Using Optimal Sensing

Abstract

The raising process has been widely used in manufacturing fabric productions. After raising the surface of the fabric, productions are covered with a fluff layer. The quality of the fabric surface is often valuated by the fluffing type. In order to objectively assess the fluff quality of the fabric surface, an optimal sensing method is proposed in this paper. The fluff contour image was firstly collected by the light-cut imaging device. Then, the fluff region was segmented by the adaptive image segmentation method, the contour coordinates of the fabric were extracted using the freeman chain code and constructed in the form of the binary image. Lastly, a back-propagation neural network (BPNN) was used to learn the relationship between the contour coordinates and the fluff quality. On this basis, a practical fabric fluff detection platform was developed based on the optimal sensing technique. Experimental tests were conducted to evaluate the performance of the proposed method in detecting the fluff quality with four different colours and different fluffing processes. Furthermore, the actual fabric inspection was carried out. The detection correct rate can reach 94.17 %, which can meet the practical production requirement.

Keywords

• fabric manufacturing
• machine vision
• artificial intelligence
• optical imaging