摘要
为了提高自然环境下茄子采摘机器人的采摘成功率并缩短采摘周期,对茄子采摘过程的目标识别进行了研究。自然环境下茄子的生长状况较为复杂,枝叶及果实间相互遮挡的情况降低了对目标识别的成功率。针对这一问题,提出了一种基于霍夫变换算法,通过在投票阶段引入条件概率,构建概率检测模型,采用基于最大后验概率的贪婪算法求解概率模型,多次迭代局部最优值确定目标,从而完成对茄子目标的识别。结果表明,对无遮挡及存在遮挡情况的茄子目标的识别正确率都在89%以上,相对于传统识别方法,该方法识别正确率大大提高,且具有较好的抗噪能力,可为茄子采摘机器人的后续开发提供参考。
In order to improve the success rate and shorten the picking cycle of eggplant under natural environment,the target identification of eggplant in the picking process was studied.Under natural environment,the growth condition of eggplant is more complicated,and a large number of branches,leaves and fruits occlude each other,which reduces the success rate of target identification. In view of this problem, this paper proposed a Hough transform algorithm based on the introduction of conditional probability in the voting phase to build a probabilistic model.A greedy algorithm based on the maximum posteriori probability was used in the voting phase,and the local optimum was iterated multiple times and the pictures were updated to determine the target.The experimental results show that the correct detection rate of eggplant target with no occlusion and occlusion was above 89%.Compared with the traditional recognition method,the accuracy of our method was improved greatly,and it had good anti-noise ability,which could provide reference for the subsequent development of eggplant’s picking robot.
作者
王维强
付斌
WANG Wei-qiang;FU Bin(Harbin University of Commerce,Harbin,Heilongjiang 150028)
出处
《安徽农业科学》
CAS
2019年第18期224-227,共4页
Journal of Anhui Agricultural Sciences
基金
黑龙江省自然科学基金项目(F2016007)
关键词
自然环境
霍夫变换
条件概率
贪婪算法
Natural environment
Hough transform
Conditional probability
Greedy algorithm
