摘要
可乐瓶在灌装前必须经过多种指标检验.针对实际工程需要,设计并实现了一套操作简单,高识别率的空瓶瓶口图像检测系统;图像经过初步处理后,去除噪声,边缘检测,将所要处理的图像有效部分与背景区分开来,如此可大大减少后续算法的计算量,从而有利于提高处理速度,较好地解决了,实时性与准确性之间的矛盾;采用基于最小二乘法的椭圆拟合方法对轮廓进行拟合,确定出椭圆的参数,并对内外圆的圆心取平均值,定为共同的圆心;经实践证明分辨力小于0.5mm,正确识别率达到99.9%以上,满足实际生产要求。
The quality of Coke bottles should be inspected before the filling. A empty bottle image inspection system with easy operation and high discrimination rate was designed. By preliminary processing, noise elimination, edge detection, the image of mouth of Coke bottle can be picked up from background image. For fitting the contour, the least square method is applied to get the parameters of the ellipes of bottle mouths. The resolution of the system is less than 0. 5mm and discrimination can reach 99. 9%. Our system will have a good application.
出处
《计算机测量与控制》
CSCD
2008年第4期449-451,共3页
Computer Measurement &Control
关键词
瓶口检测
去除噪声
边缘检测
椭圆拟合
empty bottle inspection
remove noise
edge detection
ellipse fitting
