树干曲率的自动测量是智能立木整枝机作业的重要步骤.该文针对立木图像的特点,提出了一种基于计算机视觉的立木树干曲率自动测量方法.该方法包含4个主要步骤:利用阈值分割对树干图像进行提取,得到图像的粗分割;在中值滤波的基础上,利用拉普拉斯算子对图像进行边缘位置点的检测,从而得到准确而连续的边缘;利用数学形态学方法对树干边缘进行细化处理;以道格拉斯一普克法提取树干曲线上的特征点,对其进行曲线拟合并求得树干曲率.实验结果表明,该方法能比较准确地识别树干的弯曲度.
For the automatic pruning of the standing trees by pruning robot, the automatic measuring of trunk curvature of standing tree is a key step. According to the characteristics of digital image of standing tree, this paper puts forward a computer vision-based automatic measuring method for trunk curvature of standing tree. The method includes four steps: the tree trunk image is detected by threshold segmentation; the edge of tree trunk is detected by Laplaeian operator based on the tree trunk image filtered by median filtering method; the edge image of tree trunk is thinned based on the mathematical morphology; the last step is to fit the curve and compute the tree trunk curvature automatically. The experimental results indicate that the method is available for getting accurate tree trunk curvature.