针对线束连接器上印刷字符检测智能化程度不高的情况,提出了一种基于机器视觉的检测方法.首先,利用一阶导数获得印刷字符边缘点,并将边缘点拟合为线,由此定位字符的位置,并对字符的角度和位置进行比较;然后,对合格的字符分割、归一化,并以整个字符图像作为输入特征;最后,基于字符的特征,建立了用于印刷字符识别的 BP 神经网络,并对其进行了改进.实验证明:基于机器视觉的连接器印刷字符单个字符识别准确率为99.1%,检测时间为95 ms,具有较高的识别准确率和速度.%Considering the intelligent of printed characters inspection is not high for wire harness con-nection,a method based on machine vision was proposed.Firstly,the first-order derivative was used to obtain edge points of printed characters,and the edge points were fitted into a line to locate the position of characters,then,the angle and position of characters were compared.Secondly,the character was segmented,normalized,after that,a whole character image was selected as an input feature.Finally,a back-propagation neuron network was established and improved for the printed characters classification based on above feature.Experiments show that the printed characters inspection method based on ma-chine vision has high recognition accuracy and speed,of which the recognition accuracy is 9 9.1 % and the average consuming time is 95ms for single character.
展开▼
