基于机器视觉的条烟分类识别与定位研究

Research on the classification and location of cigarette based on Halcon

针对人力分拣条烟速度慢、易出错等问题,提出利用机器视觉实现对条烟的分类识别与定位。首先对图像进行预处理,在图像像素范围内辨别条烟像素所占区域;其次根据图像算子计算条烟的边缘区域,得到图像中条烟的方位向量;再次利用金字塔搜索策略寻找与图像条烟边缘匹配的模板;最后采用支持向量机的数据分类功能,实现对条烟的分类。实验表明,该方法可准确识别条烟并对其进行精确定位,对于提高工作效率具有重要意义。

Aiming at the progress of manpower sorting cigarettes, such as low speed, more error-prone, and so on, it proposes a method to realize classification and recognition of cigarette based on machine vision. It takes the pro-process images to separate the area containing cigarette from the image, conducts the edge detection and tracking, obtains the position and direction of the edge of a cigarette which lays the foundation for the creation of cigarette＇s template. Once more, it adopts the pyramid search strategy based on template matching of image edges to find templates in the image to narrow the search area, achieves the position of cigarette on image. It analyzes the classification method based on support vector machine, realizes the classification of cigarette. The experiments show that the method can accurately classify, recognize and locate cigarette, be of great significance to improve the work efficiency.