基于机器视觉的枸杞枝条识别方法研究

Research on Recognition Approach of Chinese wolfberry Based on Machine Vision

摘要:

目标枝条的识别定位是实现枸杞果实自动化采摘的前提,为了实现枸杞的自动采摘,提出了一种新的能实现首端定位的枝条识别方法。首先,通过对枸杞枝条特征进行分析,实现了枝条轮廓和直径等特征的提取。然后,对部分非合理性间断的枝条进行了修复；对于果实将枝条首端完全遮挡的情况,采取最小二乘法拟合实现了枝条识别。最后,提出了质心差值法来获取枝条的首端坐标,从而实现了整条枝条的识别和定位。通过用VC编程进行识别算法的实验验证,实验结果表明该算法能较好地实现对枸杞枝条的识别和定位。

Abstract:

The identification and positioning of the target branch is the prerequisite for wolfberry automatic picking. In order to realize the automatic harvest of Chinese wolfberry, a new identification approach is researched which can realize the head end positioning of Chinese wolfberry branches. First of all, the contour and diameter of the branches are extracted through analyzing the characteristics of branches. Then, the partial unreasonable disconnected branches are restored, and the least square fitting method is used to achieve branch location for the condition that the head end of the branch is covered by fruits. Finally, a kind of centroid difference value approach is presented to obtain the coordinate of head end of the branches, thus the identification and location are implemented for the whole branch. The algorithm has been programmed with VC, and the experimental results show that the algorithm can realize the position and recognition of branches of Chinese wolfberry.