基于图像处理的纱线条干均匀度检测

Yarn strand evenness detection based on image processing

针对图像处理在纱线条干均匀度检测中出现的纱线图像在采集时发生光照不均以及倾斜会影响最终结果的问题,研究基于同态滤波以及Hough变换对纱线条干均匀度进行检测的方法。首先对光照不均现象用同态滤波方法将图像光照分量和反射分量分离开,对光照分量加以抑制来突出反射分量,从而消除光照不均现象,突出条干部分;其次对图像中倾斜纱线用Hough变换检测倾斜角度加以旋转校正至水平状态;最后对无噪声信号纱线图像进行图像分割提取纱线主干部分从而来计算采集到纱线的平均直径。实验表明,通过计算图像处理得到的纱线条干CV值与现有条干仪检测结果对比偏差小于4%,验证了图像处理方法的准确性,对未来形成成套的视觉检测系统有参考意义。

To address the issues of uneven lighting and tilt affecting the results in yarn strand evenness detection,a method for detecting yarn strand evenness based on homomorphic filtering and Hough transform is explored.Firstly,the homomorphic filtering method is used to separate the image into illumination and reflectance components to suppress the illumination component and highlight the reflectance component,thereby eliminating uneven lighting and emphasizing the strand part.Secondly,the tilt of yarn in the image is detected using the Hough transform to rotate and correct the tilt angle to a horizontal position.Finally,the main part of the yarn in noise-free signal images is extracted through image segmentation to calculate the average diameter of the collected yarn.Experimental results show that the deviation of the CV value of the yarn strand obtained through image processing compared to the results from existing strand meters is less than 4%,validating the accuracy of the image processing method and providing reference for the future development of a complete visual detection system.