基于毛羽补偿与自适应中值滤波的纱线主体图像识别算法

Image recognition algorithm based on yarn hairiness compensation and adaptive median filter

为解决纱线参数识别与计算过程中毛羽对纱线主体图像识别的干扰问题,提出了基于毛羽部分图像灰度补偿与自适应中值滤波的算法。采用R数据将影像扫描仪采集到的RGB图像二值化处理为灰度图像;然后根据毛羽形态上垂直方向灰度值的变化规律识别毛羽(白色背景黑色纱线),并进行由小及大的3层推抹补偿,同时根据毛羽像素灰度值特点识别毛羽(黑色背景白色纱线),并用背景灰度值补偿;最后将补偿后的2种图像各自进行最大窗口小于11的自适应中值滤波。MatLab仿真结果表明,该算法可较快地识别并补偿毛羽部分像素得到清晰的纱线主干图像,处理结果可达到检测精度要求。

In order to solve the influence of yarn hairiness on image recognition of yarn body in the process of yarn parameter recognition and calculation,an algorithm based on gray scale compensation and adaptive median filtering was proposed.R data were used to binarize the RGB image collected by the image scanner into a grayscale image.The image of the black yarn with white background was recognized according to the change of the vertical gray value in the feather form and then compensated by 3 layer push from small to large.The image of the white yarn with black background was recognized and the background gray value was compensated according to the gray value of the feather pixel.Each of the two compensated images were subjected adaptive median filtering(ANF)with a maximum window smaller than 11.The results of MatLab simulation show that the algorithm can recognize and compensate some pixels of yarn hairiness quickly and acquire clear main yarn image.The results can meet the requirements of accuracy.