计算机视觉在纺纱过程中棉结杂质检测研究

摘要

研究纺纱工艺中棉结杂质的准确检测.在纺纱工艺的松解工序中需要将块状的棉纤维分解成单根纤维状态,但实际分解完成的单纤维中不可避免地存在棉结杂质,当棉结杂质较小时其形态与单纤维相似,用于识别的形态特征不明显,传统的计算机视觉的检测方法是单一尺度边缘检测算法,不能将棉结杂质从形态相似的单纤维中准确检测出来.为了提高检测的准确度,提出一种改进的计算机视觉检测方法,对待检测的单纤维图像进行处理,将图像分割成多个区域并进行小波变换,计算每个变换区域的平均梯度值作为检测阈值,通过多尺度阈值判别将棉结杂质检测出来.实验表明,改进方法能够有效区分形态相似的单纤维和棉结杂质,将棉结杂质准确检测出来,具有较优的检测准确度.%An improved computer vision detection method was proposed. For processing, the image is divided into several regions and the wavelet transformation average gradient value of each transform domain is calculated as the detection threshold value in order to improve the detection accuracy. An improved computer vision detection method was proposed tol treat the detected single fiber's image and distinguish the nep impurities by the multi - scale threshold. The experiments show that the improved method can effectively distinguish single fiber and nep impurities which are morphologically similar and has better detection accuracy.