Detection algorithm of white foreign fibers based on improved two-dimensional maximum between-class variance method

The white foreign fibers detection is a difficult problem of the lint online detection. Through the analysis of the 2D gray histogram of white foreign fibers and lint, the two-dimensional Otsu algorithm was improved. This improved algorithm considered the probabilities of the counter-diagonal area in 2D gray histogram when the probabilities of objective and background was calculated, and reduced the range of threshold. The results indicated that the improved algorithm of 2D effectively enhanced the accuracy and real-time property of segmentation comparing with the one-dimensional Otsu algorithm and the fast two-dimensional Otsu algorithm. The improved algorithm has been successfully used in practical production.