草莓采摘机器人的研究:Ⅱ.基于图像的草莓重心位置和采摘点的确定

Study on strawberry harvesting robot: Ⅱ. Images based identifications of strawberry barycenter and plucking position

成熟草莓果实重心和采摘点的确定是草莓采摘机器人采摘收获作业中的关键问题。采用LRCD (lumi nanceandredcolordifference)方法分割草莓图像。在RGB色彩模型中 ,求得图像中每个像素的色差 ,在灰度图像中显示以色差值为灰度值的色差图像 ,取合适的阈值对该图像二值化 ,得到分割后的草莓图像 ;提取分割后草莓图像的几何特征 ,从而确定草莓的重心位置和采摘点。采用本文所述的机器人视觉系统 (CCD成像像素 75 3× 5 82 ,像素中心距 10 μm× 10 μm)的试验结果表明 ,其采摘点位置误差 <3mm。利用LRCD方法能够很好的将成熟草莓与背景分离 ,通过提取分割后草莓图像的几何信息 ,可确定草莓重心和采摘点的位置。

During the strawberry collection process by robot, the crucial element to ensure the picking precision is the determinations of the ripe strawberry barycenters and the plucking positions. The LRCD (luminance and red color difference) method was adopted to dissect the image of strawberry. The Chromatism of each pixel in the RGB color image was calculated and the result showed in a gray image. The segmented image was gained by translating the gray image to a binary image after selecting an appropriate threshold. The barycenter of strawberry and the plucking position were ascertained by extracting geometry feature of the segmented strawberry image. The result reveals that the eccentricity of the plucking position is not more than 3mm in the robot vision system(CCD pixel 753×582,distance between pixels 10μm×10μm) in this paper. Ripe strawberry can be separated easily from the complex background by the LRCD method. The method is valid for pick-up of red object from a complex background.