摘要
为实现视觉导航的精确性和鲁棒性,研究了温室复杂环境机器人视觉导航路径的识别方法。以温室内西红柿苗垄为研究对象,在地膜、光线和阴影等复杂环境对植物识别的影响下,用Lab色彩空间将绿色植物从背景中分离出来;用基于权重因子的阈值分割算法代替常用的阈值分割算法;用改进的Hough变换的导航线提取方法处理有杂草干扰的作物垄。试验证明,该方法对复杂温室环境下作物垄导航线的提取有较好的适应性,而且算法简单,能够满足实时性的要求。
In order to implement the accuracy and robust of vision navigation,the method of tomato seedlings ribbing lane detection for vision navigation in greenhouse was studied. Plants in the images were successfully recognized from complex background with plastic film,ray of light and shadow in a Lab color space. A threshold segmentation based on proportion coefficient was proposed instead of a routine threshold segmentation method to obtain binary image. A crop ribbing recognition method based on ameliorated Hough was proposed to eliminate the influence of weeds. Experiment results prove the method is effective to get the target guidance path in the complex greenhouse environments and is simple and meeting real-time need.
出处
《农机化研究》
北大核心
2010年第6期83-86,共4页
Journal of Agricultural Mechanization Research
关键词
视觉导航
路径识别
HOUGH变换
vision navigation
path recognition
Hough transformation
