A new approach to yarn neps detection based on mathematical morphology

The quantity and size of yam neps are important ingredients for assessing yarn quality. Affected by noise and restricted to the precision of sensor, yam neps detection and classification are still difficult problems. In this paper, a new morphological approach to yam neps detection is proposed, which is characterized by constructing a trapeziform structural detection template from dilating and eroding the yam linear density signal with various structuring elements to detect yam neps. Finally the neps are classified according to their unitary linear density. Results of simulation based on real yam data are shown that the effectiveness and veracity of this approach as well as strong Robustness.