基于改进亚像素边缘提取的一种异性纤维检测方法

A Foreign Fiber Detection Method Based on an Improved Sub Pixel Edge Detection

为了提高异性纤维检测的时效、精确性,降低误检率,论文提出一种基于改进的亚像素边缘检测技术上的异性纤维检测方法,首先以多结构元素的改进的形态学算子对采集的图像进行像素级边缘提取,然后利用三邻域的非极大值抑制方法抑制初步提取时膨胀的边缘以及去除误检测的小范围棉花边缘,最后使用基于Zernike矩的亚像素边缘检测方法进行亚像素级细化检测。通过实验验证,文中的方法对各种常见噪声都具有抗噪滤噪能力强、计算速度快等优点,能够快速准确地识别高速棉流中的异性纤维,满足生产中异性纤维拣出的性能需求。

To improve the accuracy of detecting foreign fibers and the time efficiency,reduce the false detection rate,a foreign fiber detection method based on improved sub pixel edge detection is proposed.At first the captured images from camera is preprocessed through an improved morphology operator,acquire preliminary marginal texture detailed images on the pixel level.Then three neighborhood non-maximum suppression method is used to remove small area cotton edge with detection errors.Finally thinning detection is conducted by sub pixel edge detection method based on Zernike moment.The experiment proves that this method has powerful noise resistance and denoising ability,fast calculation and other advantages,it can quickly and accurately identify foreign fiber in high-speed cotton flow and meet the performance requirements in actual production.