基于图像序列差分法的冷轧薄板板形识别

为了解决板形识别中存在的实时性差、准确度低、维修成本高的问题,提出了一种计算机视觉技术的板形缺陷识别.采用图像序列差分法提取图像中的运动对象信息,再利用最小误差阈值分割方法进行图像二值化,生成板形缺陷二值图像,通过统计窗口区域中像素为1的数目来判定板形是否有缺陷.提出了基于图像序列差分法识别板形缺陷的算法,并且实现了该方法的硬件和软件设计.实验结果表明,此方法可以快速准确地识别出常见的板形缺陷.

Cold-rolled Strip Plate-profile Recognition Based on Image Sequence Difference

A computer vision technology for flatness defect recognition is presented to solve the problems in the real-time identification of the poor,low accuracy,high cost of maintenance in shape recognition.Image sequence difference is adopted to make the campaign target information in image prominent,the smallest error threshold segmentation method is used to make image binarized,shape defects binary image is generated,the number of flatness in the Census and Statistics window for a pixel in the region is added up to determine whether there is a defect in shape.Flatness Defect identification algorithm based on image sequence difference as well as the methods to achieve its hardware and software design is proposed.Field experiments show that this method can quickly and accurately identify the common shape defects.