摘要
针对透明液体生产线上,人工检测效率和准确率低的不足,提出了基于加窗小波变换的杂质检测方法,该方法分别在图像预处理和检测算法上进行了研究分析;获取的图像采用高斯同态滤波来提高图像的对比度,利用基于加窗的小波变换来提取图像中杂质的边缘特征点,然后采用迭代阈值分割方法来分割目标和背景并通过显示的结果中有无亮色区域来判断透明液体中是否含有杂质,若显示有亮色区域,则判此透明液体中含有杂质;实验结果表明,此方法比现有的其他检测算法有较好的检测率和准确率,而且本算法较好的判断出了细小而低对比度的杂质,也避免了漏检微弱边缘的问题。
For the production line of transparent liquid . in view of the low efficiency and low accuracy of artificial detection, impurity detection method of wavelet transform based on adding window is put forward. In this method , the image preprocessing and detection algo rithm were studied and analyzed. And theGaussian homomorphic filtering was adopted to improve the contrast of the image , the edge lea ture points of impurities in the image are extracted by using the wavelet transform based on window, then using the iterative threshold seg mentation method adopted to segment the object and background, and by showing the results have bright region or not to judge whether transparent liquid containing impurities, revealed the bright region, then judged this transparent liquid contains impurities. The results show that this method has better detection rate and accuracy rate other than existing algorithms for detection and this method can distinguish ed the small and low contrast impurities, and removed interference better than other existing algorithms for detecting , and also this method avoid the problem of undetected weak edge.
出处
《计算机测量与控制》
2016年第10期57-60,共4页
Computer Measurement &Control
关键词
杂质检测
小波变换
特征提取
阈值分割
impurity delection
wavelet transform
feature extraction
threshold segmentation
作者简介
王一丁（1967-）．男．辽宁沈阳人．教授，博士，主要从事模式识别与图像处理、工业无损捡测、无线电定位技术方向的研究.
