C# 创建KD树的程序

最新推荐文章于 2025-05-18 22:07:43 发布

原创最新推荐文章于 2025-05-18 22:07:43 发布 · 1.3k 阅读

7 ·

CC 4.0 BY-SA版权

机器学习同时被 2 个专栏收录

8 篇文章

订阅专栏

4 篇文章

订阅专栏

实现了KD平衡树的程序，由于MATLAB实现需要用到引用类型或者采用循环实现(见MATLAB的KDTreeSearcher.m),因此采用C#来实现

using System;
using System.Collections.Generic;
using System.Linq;


namespace KNNSearch
{
    /// 
    /// Description of KNN.
    /// 
    public class KNN
    {
        /// 
        /// 叶子节点点的个数
        /// 
        private int leafnum = 30;
        /// 
        /// 待分类数据
        /// 
        private List rawData;
        /// 
        /// 生成原始数据
        /// 
        private void GeneralRawData()
        {
            if (rawData == null)
            {
                rawData = new List();
            }
            else
            {
                rawData.Clear();
            }
            Random r = new Random();
            for (int i = 0; i < 500; i++)
            {
                rawData.Add(new Point() { X = r.NextDouble(), Y = r.NextDouble(), Z = r.NextDouble() });
            }

        }

        /// 
        /// 创建KD树
        /// 
        /// 
        /// 
        private Node CreateKDTree(List data)
        {
            // 创建根节点
            Node root = new Node();
            // 添加当前节点数据
            root.nodeData = data;
            // 如果节点的数据数量小于叶子节点的数量限制，则当前节点为叶子节点
            if (data.Count <= leafnum)
            {
                root.leftNode = null;
                root.rightNode = null;
                root.point = double.NaN;
                root.splitaxis = -1;
                return root;
            }
            // 找到分割轴
            int splitAxis = GetSplitAxis(data);
            // 分割数据
            Tuple, List> dataSplit = GetSplitNum(data, splitAxis);
            root.splitaxis = splitAxis;
            root.point = dataSplit.Item1;

            root.leftNode = CreateKDTree(dataSplit.Item2);
            root.rightNode = CreateKDTree(dataSplit.Item3);
            return root;
        }

        private Tuple, List> GetSplitNum(List data, int splitAxis)
        {
            // 对数据按照第splitAxis排序
            var data0 = splitAxis == 0 ? (data.OrderBy(x => x.X)).ToList() : 
                (splitAxis == 1 ? (data.OrderBy(x => x.Y)).ToList() : 
                (data.OrderBy(x => x.Z)).ToList());
            int half = data0.Count / 2;
            List leftdata = new List();
            List rightdata = new List();
            for (int i = 0; i < data0.Count; i++)
            {
                if (i <= half)
                {
                    leftdata.Add(data0[i]);
                }
                else
                {
                    rightdata.Add(data0[i]);
                }
            }
            double splitnum = splitAxis == 0 ? data[half].X + data[half + 1].X :
                (splitAxis == 1 ? data[half].Y + data[half + 1].Y :
                data[half].Z + data[half + 1].Z);
            return new Tuple, List>(splitnum / 2, leftdata, rightdata);
        }
        /// 
        /// 获取分割轴编号
        /// 
        /// 
        /// 
        private int GetSplitAxis(List data)
        {
            // 设定数据范围最大的轴作为分割轴（也有其他的方式，如方差，或者轮流的方式）
            var xData = data.Select(item => item.X);
            var yData = data.Select(item => item.Y);
            var zData = data.Select(item => item.Z);
            List ranges = new List();
            ranges.Add(xData.Max() - xData.Min());
            ranges.Add(yData.Max() - yData.Min());
            ranges.Add(zData.Max() - zData.Min());
            var sorted = ranges.Select((x, i) => new KeyValuePair(x, i)).OrderByDescending(x => x.Key).ToList();
            return sorted.Select(x => x.Value).ToList()[0]; ;
        }

        public KNN()
        {
            GeneralRawData();
            Node node = CreateKDTree(rawData);
        }
    }

    /// 
    /// Description of Node.
    /// 
    public class Node
    {
        /// 
        /// 切分的阈值点
        /// 
        public double point;
        /// 
        /// 左节点
        /// 
        public Node leftNode;
        /// 
        /// 右节点
        /// 
        public Node rightNode;
        /// 
        /// 节点包含的数据
        /// 
        public List nodeData;
        /// 
        /// 分割轴
        /// 
        public int splitaxis;
        public Node()
        {
        }
    }

    public class Point
    {
        public double X;
        public double Y;
        public double Z;
    }
}