The determination of the twist level of the Chenille yarn using novel image processing methods: Extraction of axial grey-level characteristic and multi-step gradient based thresholding

This paper presents, a new and fast image analysis based method, namely extraction of axial grey-level characteristic (EAGLC), for directly determining the dimensions of a local structural texture, as observed on the axis of the Chenille yarn. The EAGLC signal, obtained from the texture composed by the helically wrapped core yarns, allows identifying the twist level and its variations. Additionally, a novel indirect method, namely multi-step gradient based thresholding (MSGBT), including parametric equation of the elliptical cross-section helical yarn is developed to obtain the twist level using algorithmic thresholding process. Hough Transform method (HT) allows to verify the twist and twist orientation values obtained by the MSGBT method. Consequently, the EAGLC method can be used for the weak local texture compared with grey-level projection method used for strong periodic patterns as are the case in warp and weft yarns in fabric. A yarn inspection system, including optoelectronic sensor, is developed in conjunction with computer vision unit to assess the signal changes corresponding to the image regions where the twist level is determined.