花椒果实的目标识别与定位

Target recognition and location of Zanthoxylum bungeanum fruit

为解决花椒背景复杂、光暗条件变化、果实串重叠等因素给智能采摘机器人作业带来的定位识别不准确的问题,提出融合多种算法的识别定位方法,并基于形心偏差法与惯性主轴算法提出了一种交廓等腰三角形算法模型。实验结果表明,针对所选取的不同光照条件下无遮挡与遮挡的花椒图像,计算采摘点准确率达到79%和62%,平均定位时间为0.2655 s和0.3126 s。可为花椒采摘机器人的研究提供理论依据。

In order to solve the inaccurate positioning and recognition caused by the complex background of Zanthoxylum bungeanum,the change of light and dark conditions and the overlap of fruit strings.In this paper,a recognition and positioning method integrating multiple algorithms is proposed.Based on the centroid deviation method and inertial principal axis algorithm,an algorithm model of intersection isosceles triangle is proposed.The experimental results show that the accuracy of picking points is 79%and 62%,and the average positioning time is 0.2655 s and 0.3126 s.Provid theoretical basis for the research of pepper picking robots.