基于NMI的织物疵点粗定位算法研究

The Research of a Algorithm for Fabric Defects Coarse Positioning Based on the NMI

针对几种常见的含疵点的织物图像,本文提出了一种基于NMI的疵点粗定位算法.首先,对含疵点的织物图像进行去噪和二值化;然后,借鉴NMI提取特征的思想,针对疵点的形状,对二值化后的图像进行相应的分块;最后,根据每个子块NMI的大小,粗定位出原织物图像中疵点位置.实验结果表明,该算法能够获得较好的粗定位效果.

A new algorithm for fabric defects coarse positioning based on the NMI was proposed for several common defects in fabric images. Firstly, the fabric image was pre-processed to get the binary image. Then inspired by the idea of NMI for feature extraction, the binary image was partitioning corresponding block to the shape of defects. Lastly, locating the approximate position of defects according to the value of each sub-block NMI. The result of experiment shows that the new algorithm is effective for locating the defects.