基于全局阈值迭代的苹果图像分割计数方法

Counting Method of Global Threshold Iterative Algorithm for Apple Images Based on Global Threshold Iterative

针对苹果分拣生产线检测分级过程中面临的多个苹果堆叠而产生的计数精度不高的问题,提出了一种全局阈值迭代算法,在RGB图像空间利用抓取的彩色苹果图像R值矩阵生成灰度图,通过多次迭代计算,提取了各连通域质心的欧式距离特征,并针对堆叠苹果的粘连连通域进行了堆叠质心的剔除,进行了三次迭代计算之后,最终确定了图像中苹果的数量,计数精度达到90%,该研究可为苹果分拣生产线的自动化设计提供技术支持。

A global threshold iterative algorithm was proposed for multiple apple stacking in the process of detection and classification in apple production line.R value matrix of apple image in RGB image space was used to segment the adhesive apples.The Euclidean distance characteristics of the centroid of each connected domain were extracted by multiple iterative calculation and the repeated centroid was eliminated.After three iterations,the number of apples was determined and the precision of the algorithm was 90%.The research can provide technical support for automatic design of apple selection line .