名优绿茶嫩芽识别与定位方法研究  

Approach for recognition and positioning of tea shoots

作　　者：袁加红[1] 张中正[1] 朱德泉[1] 侯建[1] 朱永宽[1] 裴韶 刘鹏[1] 

机构地区：[1]安徽农业大学工学院,合肥230036

出　　处：《安徽农业大学学报》2016年第5期676-681,共6页Journal of Anhui Agricultural University

基　　金：安徽农业大学大学生创新基金(XJDC2014308)资助

摘　　要：由于人工采摘名优绿茶效率太低,严重制约了名优绿茶产业的发展。研究具有选择性且能快速准确采摘名优绿茶的智能系统迫在眉睫。而嫩芽识别和定位则是实现名优绿茶机械采摘的前提,因此,开展了名优绿茶嫩芽识别与定位方法的研究。基于RGB颜色空间采用三基色以及组合因子对图像进行灰度化处理,利用维纳滤波和梯度增强技术对处理后的图像进行滤波去噪,采用大津法和迭代法分割图像获取二值图像;借助MATLAB软件对相机进行标定处理获取各项内外参数,然后采用质心法对目标进行定位。结果表明,利用G-B索引因子结合迭代法能较大程度地识别出嫩芽目标,采用张氏标定法能在误差要求范围内对目标质心进行定位,且实时性较高。Nowadays the low efficiency of manual tea plucking is the main constraint to the development of tea industry. An intelligent plucking machine that can selectively pluck the tea shoots will increase the plucking efficiency. The key part of the machine is the one for recognition and positioning of tea shoots that should be plucked. Firstly, in the RGB color space, R, G and B indexes and their combination were applied to two gray proc- essing of images of the same tea tree which had been captured from different visual angles. The image enhancing method and wiener filtering were conducted. The Otsu and interative methods were used to segment the images and the centroid points of the shoot regions could be obtained by feature extraction. The intrinsic and extrinsic parameters were determined by calibrating the camera in MATLAB software and the coordinates of the centroid points could be calculated, which would contribute to positioning of the target. The results showed that the shoots in the images could be recognized well and rapidly by applying G-B index with interative method. The centroid points could be positioned on the basis of the intrinsic and extrinsic parameters of the camera within the allowed error range.

关 键 词：G-B因子 目标识别 摄像机标定 深度恢复 

分 类 号：S126[农业科学—农业基础科学] S571.1