【摘要】 提出了一种基于机器视觉的鞋底运动轨迹识别新方法,首先,自主开发了鞋机自动涂胶系统。引入视觉检测技术,采集鞋底图像,对采集到图像进行中值滤波及灰度增强处理。在此基础上,对遗传算法进行深入研究,研究了基于遗传算法的图像阈值分割方法。结果表明,与传统的ostu法相比,新方法能较好地去除鞋底周边的干扰,准确提取鞋底信息。最后对分割后的图像进行了数学形态学及边缘检测处理,以提取鞋底轮廓边缘,最终的图像误差为±0.03mm,为下一步鞋底涂胶轨迹的自动生成奠定基础。更多还原

【Abstract】 A new mehod based on computer vision for tread motion trajectory recognition is proposed in this paper. Firstly,the automatic shoe gumming system is developped. The vision sensor is used to capture the tread images in real time. Then the median filter and grey enhance processing are performed to the images. On this basis,the genetic algorithm has been deeply researched,and used to the image threshold segmentation. Comapared with the traditional ostu method,the new method can remove the disturbance around the tread effectively,and gain the tread information accurately. In the end,the mathematical morphology and edge detection operations are performed to the segmented images. So the contour edge of the tread can be extracted and the final image error is ±0.03 mm,which make a basis for the generation of tread gumming trajectory.更多还原