摘要
针对常见的布匹瑕疵检测方法对破洞、油污检测不敏感的缺点,提出了一种将图像边缘检测和小波分析相结合的检测方法。该方法根据瑕疵边缘变化的差异,通过改进的Canny算子提取布匹边缘特征,识别出油污;选用最佳小波对非油污图像分别在水平、垂直方向进行三层分解,提取水平和垂直方向高频图像的特征值(能量、方差和极差),得到图像纹理频谱相应特征值的分布情况;对特征值归一化并设置相应的阈值,即可得到不同的特征向量,从而实现对断经、断纬和破洞的实时检测。实验表明,该方法的检测速度快,准确率高,可以满足检测要求。
For the common fabric defect detection method is not sensitive to distinguish between holes and oil, this paper proposes a method based on a combination of image edge detection and wavelet analysis. In terms of the differences of the changes in marginal defects, oil can be identified by extracting edge features with improved Canny operator. Hereafter, respectively decompose the image in the horizontal and vertical direction in the scale of 3 with adaptive wavelet to extract horizontal and vertical high-frequency images＇ characteristic value： energy, variance and range and get the distribution of related features value on the image texture spectrum. Finally, normalize the image and set up corresponding threshold so as to get a set of characteristic vector through which end out,thread out and torn can be defected. Experiments showed that the method is quickly, high accuracy and can meet the detection requirements
出处
《江南大学学报（自然科学版）》
CAS
2011年第6期637-641,共5页
Joural of Jiangnan University (Natural Science Edition)
基金
国家自然科学基金项目(60804013)
江南大学大学生创新训练计划项目(1003046)
关键词
瑕疵检测
边缘检测
小波分析
特征值
fabric defect detection, edge detection, wavelet analysis, characteristic value
