摘要
由于田间路径通常呈现的边界模糊和形状不规则特点,常用的道路导航点提取算法应用在农业环境会出现偏差过大的问题.本文设计了一种基于圆环模型的适应度函数,对遗传算法进行了改进,提出了适应度尺度变换和种群初始化的自适应策略,避免了陷入局部最优解和收敛速度慢的情况.经仿真,相比于坐标中点法提取导航点,改进后遗传算法的辨识误差降低到了3.52%以下,能够有效地解决环境噪声干扰下的田间路径导航点提取问题.
Due to the boundary blur and irregular shape of the field path,the commonly used road navigation point extraction algorithm may have excessive deviation.In this paper,a fitness function based on the ring model is designed and the genetic algorithm is improved.An adaptive strategy for fitness scale transformation and population initialization is proposed and it avoids the situation of falling into the local optimal solution and slow convergence.Through simulation,compared with the coordinate midpoint extraction of navigation points,the identification error of the improved genetic algorithm is reduced to below 3.52%.The improved algorithm can effectively solve the problem of field path navigation point extraction under environmental noise interference.
作者
孙涵
江剑
丁良华
齐昕雨
SUN Han;JIANG Jian;DING Lianghua;QI Xinyu(School of Mechanical and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China;Inner Mongolia North Heavy Industries Group Co. Ltd. Technology Center Product Research Institute, Baotou 014000, China)
出处
《测试技术学报》
2020年第3期227-232,共6页
Journal of Test and Measurement Technology
关键词
机器视觉
田间路径
改进遗传算法
导航点
machine vision
field path
improved genetic algorithm
navigation point
