树叶长宽比及边缘曲率的提取

THE EXTRACTION OF LEAVES’ ASPECT RATIO AND BOUNDARY CURVATURE

  • 摘要: 树叶的轮廓特征主要包括长宽比、面积、边缘曲率和斜率等,这些特征参数是树木分类及识别的关键.为此,利用数学形态学与像素跟踪相结合的算法分离叶柄,然后利用霍特林(Hotelling)变换获取树叶的最小外接矩形,进而得到树叶的长宽比. 基于有限差分算法得到2种不同形态树叶的边缘曲率和树叶关键部位的特征斜率.

     

    Abstract: The contour features of a leaf mainly include the aspect ratio, area, boundary curvature, and slope, which are very important to classify and recognize the tree. The mathematical morphology combined with pixel tracking algorithm is used to separate petiole of the leaves. Using the Hotelling transform, the minimum enclosing rectangles and the aspect ratios of leaves are obtained. Based on the finite-difference algorithm, the boundary slopes and curvatures of the key parts of the different leaves are also obtained.

     

/

返回文章
返回