基于数学形态学的棒材自动计数方法

Method for Automatic Counting of Steel Bars Based on Mathematical Morphology

针对图像中棒材存在粘连,从而造成误计数的问题,研究了一种基于数学形态学的棒材自动计数方案,以实现对棒材实时、精确的计数。该方案针对棒材类计数无需考虑被测物体截面真实形状的特点,将每根棒材处理成单个像素,避免了粘连对计数的影响。为有效克服光照、噪声的干扰以及棒材快速运动造成的图像模糊,该方案首先对采集到的原始棒材图像进行预处理,然后采用腐蚀、细化算法初步实现了棒材粘连的分离,最后采用收缩与检测直线端点算法,实现了单个像素点与原始图像棒材的一一对应,为后续计数奠定了基础。实验证明,该方法不仅能够保证棒材自动计数的精度,并且为棒材定支数包装创造了条件。

The paper proposes a method to realize the real-time and accurate counting of steel bars. Through using the method of pro cessing every bar into a single pixel, the proposed scheme can avoid the influence of the counting by adhesion. Meanwhile this method makes the true shape of object section about steel bars counting unnecessary to consider. To effectively overcome the image blurring caused by the lighting, noise interference and the quick movement of the bar, the proposed scheme firstly proceeds image pre-process ing of the original bar images, then uses corrosion and thinning algorithm to achieve the separation of the bar adhesion, and finally uses linear shrinkage and endpoint detection algorithm to achieve the correspondence between the single pixels and the original bar images re spectively to realize the subsequent counting. Experiments demonstrate that the method can not only ensure the accuracy of automatic counting but also create conditions for the bar counts for packaging.