提出了一种基于图像连通区域面积的成捆棒材自动计数方法.该方法首先对获取的棒材图像进行灰度化、滤波和图像分割处理,然后对分割图像中的连通区域分类:小于基本面积的区域视为噪声区域,介于基本面积和阈值面积之间的区域视为一支棒材区域,大于阈值面积的区域表示多支棒材粘连区域.对于多支棒材粘连区域,采用区域面积除以标准面积得到棒材支数.最后将所有连通区域的棒材支数求和即可得到图像中全部棒材的支数.该方法应用于13幅成捆棒材图像计数实验时所得误差均小于3％.%An automatic counting method of steel bar in a bundle based on connected domain area of image was proposed. The steel bar images are first segmented after gray processing and filtering. Then connected domains in segmentation images are classified into three groups: the domains with area smaller than basic area are considered as noise, the domains with area between basic area and threshold area are representative of one steel bar, and the others are multi-steel bar, whose area are divided by standard area to get amount of steel bar. The amount of all steel bars in the image is gotten through summing up amount of steel bars in each segmentation image. The experiments with 13 images showed that all the counting errors are smaller than 3% by use of the method.
展开▼
