基于数学形态学的烟支计数识别

Cigarette counting recognition based on mathematical morphology

根据烟支的几何特征,设计1种基于数学形态学的烟支图像分割并计数识别方法．该方法对二值图像应用形态滤波器去噪,对单个标记区域进行形状分类并使用不同的结构元素进行开运算,在腐蚀后膨胀前进行单烟支判定,达到分类处理的效果．在整个形态处理过程中引入形态编码概念,从而将形态学运算转换为字节中的位运算,同时将二值图中的区域转换为整数集合,作为1种区域描绘应用在单烟支判定中．

With cigarette geometrical features, an algorithm for cigarette image segmentation and counting based on mathematics morphology is designed. It applies morphological filter to suppress various noise for binary image, classify the geometrical features for single tagged region, and use different structure elements to open the region. Single cigarette can be distinguished before its dilation and after its erosion. In the whole morphological processing, feature coding is introduced to turn morphological operations into bit-level binary operation units. The regions in binary image are converted into integer collection as a region descriptor application on single cigarette recognition.