Abstract: The growth status of roots can reflect the region’s climate and soil characteristics, and whether the root is flourishing or not can response the whole plant’s health. Research on the ecological parameter of root is the foundation of the root’s growth condition. Manual processing of minirhizotron images is time-consuming and strenuously, this paper puts forward a kind of automatic segmentation method in minirhizotron images. It extracts centerlines of minirhizotron images based on ridge detection, then computes the directional information of each pixel in centerlines, uses the directional local contrast method to detect pixels on either side of the centerline. Experimental results show that compared with the traditional method, this method is effective.

Key words: image segmentation, minirhizotron images, centerline, directional local contrast

摘要： 植物根系的生长状况可以反映该地区的气候以及土壤特性，根系的发达与否也能够反应整株植物的健康情况。研究根系的生态参数是研究根系生长状况的基础，目前对于微根窗图像的处理都是通过人工手动方式实现的，费时费力。提出了一种微根窗图像根系的自动分割方法，使用基于脊线检测的方法获取微根窗图像中根系的中心线，计算中心线上所有像素点的方向，采用改进的定向局部对比度方法检测出中心线两侧的根系像素得到分割后的图像。实验结果表明，与传统分割方法相比，方法具有良好的分割效果。

关键词: 图像分割, 微根窗图像, 中心线, 定向局部对比度