摘要
为解决织物疵点自动检测算法对不同种类疵点检测的鲁棒性问题以及算法的实时性问题,提出针对疵点图像的改进的引导滤波算法及其疵点检测算法。实验表明:引导滤波在保持边缘的同时速度最快,该改进算法使用高斯滤波后图像作为输入图像,原图像为引导图像,在保持细节的同时锐化并保持边缘,一张图片滤波时间为26.2 ms。该检测算法使用改进后引导滤波,再进行光照校正等处理得到疵点检出图片。实验表明,该检测算法疵点检出率为97.35%,检测时间为158.1 ms,保持检出率的同时达到实时性要求。
To solve the robust problem of fabric defect automatic detection algorithm for different kinds of defect detection and the real-time problem,it proposed for the improvement of guided filtering algorithm and defect detection algorithm based on the improved guide filter. Experiments showed that the guided filter was the fastest algorithm of edge-preserving smoothing operator,the improved algorithm using a Gaussian filter image as the input image,the original image as the guidance image,to ensure in keeping the details while sharpening and preserving edge,processing time of a piece of picture 26. 2 ms.The detection algorithm used the improved guide filter and the illumination correction and other process to obtain defect detection image. Experiments showed that the algorithm defect detection rate was 97. 35%,the processing time for 158. 1 ms,achieved the real-time requirement.
出处
《毛纺科技》
CAS
北大核心
2017年第11期70-73,共4页
Wool Textile Journal
关键词
疵点检测
图像滤波
引导滤波
鲁棒性
实时性
defect detection
image filtering
guide filter
robustness
real-time
