Pattern Segmentation of Warp Knitted Fabric Based on Improved 3D OTSU Algorithm
Aiming at the problem that the pattern of warp knitted fabric is easy to be deformed in the reshaping process,it is diffi-cult to achieve effective pattern segmentation.An improved three-dimensional maximum between-class variance(OTSU)method is proposed to segment warp knitted fabric patterns to overcome the disadvantages of traditional OTSU method for segmenting warp knitted fabric patterns,such as poor noise resistance,high computational complexity and long running time.First,denoise the warp knitted fabric image,identify and correct the noise of the image.Then,the space complexity of the three-dimensional OTSU method is reduced,and the multi-dimensional calculation is decomposed into multiple single-dimensional calculations.Finally,an improved crowd search algorithm is used to optimize the single-dimensional threshold,and the three-dimensional OTSU method that obtains the optimal threshold is used for pattern segmentation.Compared with the traditional 3D OTSU meth-od,the improved 3D OTSU method is more robust and accurate,and reduces the calculation time by 60％.
万方数据
维普
