基于光照色彩稳定性分析的温室机器人导航信息获取

Navigation Information Acquisition Based on Illumination Chromaticity Stability Analysis for Greenhouse Robot

针对温室环境下视觉导航存在光照波动影响问题,从应用可行性角度,研究了一种基于光照色彩稳定性的导航信息获取方法。分析了温室导航线图像在颜色空间的色彩特征分布,建立了光照分析模型,以提高不同光照条件下的导航线提取适应性。提出g、Cr、Cb色彩分量组合的导航线分割方法,实现了导航信息有效快速识别。采用基于导航参考点的改进Hough变换拟合导航路径,通过几何关系获得导航侧向位移和角度偏差作为导航参数。温室环境下随机拍摄150幅不同光照导航图像进行识别算法验证,导航线平均识别率为93.9%。导航实验结果表明,导航线获取方法平均耗时约95 ms,在运行速度低于1.5 m/s时最大路径跟踪误差小于6 cm,满足视觉导航实时性和准确性的要求。

For the problem of sunlight fluctuation influence on vision navigation in greenhouse environment,a navigation information acquisition method based on illumination chromaticity stability was proposed.By comparing the distribution of navigation line image in different color spaces,a light-color analysis model was established to the processes of navigation detection in variable illuminations.Combination model of g,Cr,Cb color components was used to extract guidance line rapidly and effectively.The navigation parameters including lateral deviation and angle deviation were obtained from reference navigation line fitted by an optimized Hough transformation.150 images with navigation information taken under different light conditions were tested by the recognition algorithm,which can detect navigation line with a right recognition rate of 93.9%.The navigation experimental results showed the average processing time of navigation parameters acquisition was about 95 ms,and the maximum error of path tracking was less than 6 cm with the speed lower than 1.5 m/s,which satisfied real-time requirements of vision navigation.