基于Mean-shift的工件自动提取与跟踪方法

Method of Automatic Extraction and Tracking for Workpiece Based on Mean-shift Algorithm

针对传统的Mean-shift跟踪方法在目标跟踪前需要手动截取目标对象的问题,提出了一种自动提取目标的跟踪方法。该方法在传统的Mean-shift跟踪方法基础上,融入了绝对误差和方法(SAD),通过预先识别目标对象信息并框选出来,实现自动跟踪,提高跟踪效率。采用自动涂胶机作为实验背景,模拟自动涂胶机识别并跟踪目标工件。实验结果表明,该方法通过预先识别有效的提取出目标工件并进行跟踪对准,与传统的Mean-shift方法效果一致,简化了目标跟踪前期步骤,具有较高的效率。

To the traditional Mean-shift tracking method that need to manually intercept the target object before the target tracking.A tracking method for automatically extracting targets is proposed.This method which is based on the traditional Mean-shift tracking method is added the Sum of Absolute Differences(SAD).The method can achieve the automatic tracking by identifying the target object information in advance and selecting it.And improve the tracking efficiency.We use an automatic applicator as the experimental background to simulate the automatic applicator to identify and track the target workpiece.The experimental results show,the method effectively pre-identified and extracted target track aligned workpiece.Consistent with the traditional Mean-shift method.At the same time,it simplifies the previous steps of target tracking and has high efficiency.