摘要
传统空间域疵点检测方法容易造成疵点多尺度信息的丢失。根据小波变换能有效地获取信号局部多分辨率特性,提出一种结合小波变换和图像最大熵原理的织物疵点检测方法。首先选取最优小波基对织物疵点图像进行分解,并对分解后的高频垂直及水平系数进行灰度值归一化;然后分别求出水平和垂直细节图像的最大熵及平均熵。通过比较判断出疵点类型,最后对平均熵大的细节图像进行最大熵分割,得到最终的疵点检测结果。仿真实验表明该方法对常见纬向和径向类织物疵点的检测是有效的。
Traditional defects detection method lead to information missing on muilt-scale. Based on wavelet transform can obtain local multi-resolution feature effectly, a new method of fabric defects detection combined wavelet transform with image maximum entropy is proposed. Firstly, with the optimal wavelet bases, the fabric defect images are decomposed, then to normalize gray value of the details image, the maximum and average entropy of the details image are fround. Through to compare the size and determine the type of defect, the details image are segmented with larger average entropy based on the maximum entropy, finally the test results are got. Simulation results show that this method is effective to detect common fabric defects.
出处
《科学技术与工程》
2011年第22期5446-5450,共5页
Science Technology and Engineering
关键词
小波变换
多分辨率
信息熵
疵点检测
wavelet transform multi-resolution information entropy defect detection
