Abstract

Aiming at the problem of low manual efficiency and inaccuracy in the traditional evaluation of fabric wrinkle-resistance performance relying on the eye of experts and subjective judgment, this paper proposes a fabric wrinkle grade evaluation model based on optimized regularized extreme learning machine. First, this paper extracts four image characteristics of fabric wrinkle image, including wrinkle density, gray level co-occurrence matrix, wavelet parameter standard deviation and data fusion feature, combined with the grade evaluation standard, and then obtains a complete wrinkle grade data set. Then, differential evolution is used to initialize the initial population of the Harris Hawks optimization, and the improved Harris Hawks optimization is used to optimize the input weights and hidden layer bias of the regularized extreme learning machine. Finally, this optimized regularized extreme learning machine is used to evaluate the fabric wrinkle grade. Experimental results show that the classification accuracy of the model proposed in this paper can reach 96.39%, and the proposed algorithm has no abnormal points in the analysis of the stability of the box plot.

摘要

针对传统依靠专家眼光和主观判断来评价织物抗皱性能的人工效率低、不准确的问题，提出一种基于优化正则化极限学习机的织物抗皱等级评价模型。首先，提取织物皱纹图像的4个图像特征，包括皱纹密度、灰度共生矩阵、小波参数标准差和数据融合特征，结合等级评价标准，得到完整的皱纹等级数据集。然后，差分进化用于初始化Harris Hawks优化的初始种群，改进的Harris Hawks优化用于优化正则化极限学习机的输入权重和隐藏层偏差。最后，这个优化的正则化极限学习机被用来评估织物的起皱等级。实验结果表明，本文提出的模型的分类准确率可以达到96.39%，提出的算法在箱线图稳定性分析中没有出现异常点。