摘要
提出机器视觉的方式弥补传统缺陷检测的弊端。针对目标图像模糊退化,采用中值滤波技术降噪的同时保护边缘信息;采用大津法算法进行阈值搜索;并进行直方图均衡化调整图像对比度。采用改进的Canny算子对传送带撕裂进行边缘识别提取,增加裂纹目标的可识别性。构建改进的T2FNN样本训练模型,将传送带裂纹图像的特征参数输入训练模型中,在隶属层输出模糊化的输入节点,最后通过输出层提供输入节点到输出的映射。通过验证对图像检测得到的数据结果,发现基于机器视觉基础的Canny算法在各种环境下对传送带撕裂的图像有更好的边缘检测效果,能更精准的做出判断,发出警报。
This paper presents a machine vision approach to remedy the shortcomings of traditional defect detection.According to the image degradation caused by various interference,median filtering technique is used to eliminate the noise of gray image,threshold search is performed by Otsu algorithm,and the image contrast is adjusted by Histogram equalization algorithm.The improved Canny operator is used to extract the edge of the torn belt,which increases the recognition of the cracked object.The improved T2 FNN sample training model is constructed,and the characteristic parameters such as the long axis length,the short axis length,the ratio of length to width,the Pixel area,the angle and the density of the crack image of the conveyor belt are input into the training model,finally,an input node-to-output mapping is provided through the output layer.Through the verification of the data obtained from the image detection,it is found that the Canny algorithm based on machine vision has better edge detection effect on the image with the belt torn in various environments,and can make more accurate judgments and issue alerts.
作者
张聪聪
牟莉
Zhang Congcong;Mu Li(School of Computer Science,Xi'an Polytechnic University,Xi'an 710048,China)
出处
《国外电子测量技术》
2020年第12期80-85,共6页
Foreign Electronic Measurement Technology
