python构建k-d Tree

最新推荐文章于 2023-11-12 16:18:12 发布

也许不错

最新推荐文章于 2023-11-12 16:18:12 发布

阅读量452

点赞数

分类专栏： PCL3D点云文章标签： python 概率论

本文链接：https://blog.csdn.net/qq_23363425/article/details/107796843

版权

PCL3D点云专栏收录该内容

6 篇文章 5 订阅

订阅专栏

此处不细细讲解，只做简单构造k-d Tree

选轴策略：

1.依次X，Y，Z....K轴划分

2.按方差大的轴划分，及数据分散的轴

3.按数据差值大的轴划分（本文以此策略划分）

构造k-d Tree

def get_kdtree(tree_list):
    list_lend = len(tree_list)

    if list_lend == 0:
        return None
    elif list_lend == 1:
        return {"value": tree_list[0], "left": None, "right": None}
    else:
        # 取最大差值为轴

        xlist = []
        ylist = []
        for i in tree_list:
            x, y = i
            xlist.append(x)
            ylist.append(y)
        xmax = max(xlist)
        xmin = min(xlist)
        ymax = max(ylist)
        ymin = min(ylist)
        if (xmax - xmin) >= (ymax - ymin):

            tree_list.sort(key=lambda i: i[0])

        else:
            tree_list.sort(key=lambda i: i[1])
        midNum = list_lend // 2
        left_list = tree_list[0:midNum]
        right_list = tree_list[midNum + 1:]

        return {"value": tree_list[midNum], "left": get_kdtree(left_list), "right": get_kdtree(right_list)}
  
        

if __name__ == '__main__':
    point_list = [(2, 3), (5, 4), (9, 6), (4, 7), (8, 1), (7, 2)]

    kdtree = get_kdtree(point_list)
    print(kdtree)

输出结果

{'value': (7, 2), 'left': {'value': (4, 7), 'left': {'value': (2, 3), 'left': None, 'right': None}, 'right': {'value': (5, 4), 'left': None, 'right': None}}, 'right': {'value': (9, 6), 'left': {'value': (8, 1), 'left': None, 'right': None}, 'right': None}}