基于机器视觉的枳壳自动定向方法与试验

Automatic orientation method and experiment of Fructus aurantii based on machine vision

针对人工姿态调整方法在枳壳定向切分加工过程中存在人工成本高、生产效率低等缺点,设计枳壳机器视觉自动定向调整系统,并提出相应的调整识别算法。根据枳壳表面各目标颜色特征,提出基于R分量下红绿色差法对原始图像进行分割,并利用二值图像中特征几何描述提取果梗位置特征参数和最小二乘椭圆拟合提取外形特征参数;根据两特征信息建立枳壳姿态模型用于姿态识别,定向调整装置根据视觉识别结果捕捉姿态特征并将枳壳调整到理想姿态。试验结果表明,该方法较好地识别出枳壳姿态特征,识别率为93.3%,其2次定向调整后的姿态与理想姿态最大定向夹角误差为4.2°,满足枳壳定向作业要求。该研究为枳壳定向切分加工设备的研制提供参考。

Aiming at the disadvantages of high manual cost and low production efficiency in the process of Fructus aurantii orientation and segmentation,the artificial vision adjustment system of Fructus aurantii machine vision was designed and the corresponding adjustment and recognition algorithm was proposed.According to the color characteristics of each target of the Fructus aurantii surface,the original image is segmented based on the red-green difference method under the R component,and the feature parameters are extracted by using the feature geometry description in the binary image to extract the fruit stem location feature parameters and the least squares ellipse fitting.A Fructus aurantii attitude model is established based on the two feature information for gesture recognition,and the orientation adjustment device captures the posture feature according to the visual recognition result and adjusts the Fructus aurantii to the ideal posture.The experimental results show that the method can better identify the Fructus aurantii attitude characteristics,the recognition rate is 93.3%,and the error of the maximum orientation angle between the attitude of the two orientation adjustments and the ideal attitude is 4.2°,which satisfies the requirements of the Fructus aurantii orientation operation.This study provides a reference for the development of Fructus aurantii directional cutting processing equipment.