c#实现halcon的rle编码blob分析

效果展示

实现功能

• connection
• 膨胀
• 腐蚀
• 开运算
• 闭运算
• 特征计算

核心代码

using System;
using System.Collections.Concurrent;
using System.Collections.Generic;
using System.Drawing;
using System.Linq;namespace view3d
{// 基础对象类，类似于 Halcon 的 HObjectpublic abstract class HObject{public abstract HObjectType Type { get; }public enum HObjectType{Region,XLD,Image}}// 使用 RLE (Run-Length Encoding) 编码的 HRegion 实现public class HRegion : HObject{public override HObjectType Type => HObjectType.Region;// RLE 编码数据结构: 每行存储起始列和结束列的列表private Dictionary<int, List<Interval>> _rleData = new Dictionary<int, List<Interval>>();// 边界框缓存private Rectangle _boundingBox;private bool _boundingBoxValid = false;// 用于表示一行中的连续区间private struct Interval{public int Start;public int End;public Interval(int start, int end){Start = start;End = end;}public int Length => End - Start + 1;}// 构造函数public HRegion() { }// 从位图创建区域public HRegion(Bitmap bitmap, byte threshold = 128){FromBitmap(bitmap, threshold);}// 从位图二值化并创建区域public void FromBitmap(Bitmap bitmap, byte threshold = 128,byte maxThreshold=255){_rleData.Clear();_boundingBoxValid = false;for (int y = 0; y < bitmap.Height; y++){List<Interval> intervals = new List<Interval>();bool inRegion = false;int startX = 0;for (int x = 0; x < bitmap.Width; x++){Color pixel = bitmap.GetPixel(x, y);byte intensity = (byte)((pixel.R + pixel.G + pixel.B) / 3);bool isForeground = intensity >= threshold && intensity<=maxThreshold;if (isForeground && !inRegion){// 进入区域inRegion = true;startX = x;}else if (!isForeground && inRegion){// 离开区域intervals.Add(new Interval(startX, x - 1));inRegion = false;}}// 处理行末尾仍在区域中的情况if (inRegion){intervals.Add(new Interval(startX, bitmap.Width - 1));}if (intervals.Count > 0){_rleData[y] = intervals;}}UpdateBoundingBox();}// 更新边界框private void UpdateBoundingBox(){if (_rleData.Count == 0){_boundingBox = Rectangle.Empty;_boundingBoxValid = true;return;}int minX = int.MaxValue;int maxX = int.MinValue;int minY = _rleData.Keys.Min();int maxY = _rleData.Keys.Max();foreach (var kvp in _rleData){foreach (var interval in kvp.Value){if (interval.Start < minX) minX = interval.Start;if (interval.End > maxX) maxX = interval.End;}}_boundingBox = new Rectangle(minX, minY, maxX - minX + 1, maxY - minY + 1);_boundingBoxValid = true;}// 获取边界框public Rectangle BoundingBox{get{if (!_boundingBoxValid) UpdateBoundingBox();return _boundingBox;}}// 计算区域面积public double Area(){double area = 0;foreach (var intervals in _rleData.Values){foreach (var interval in intervals){area += interval.Length;}}return area;}// 计算重心public PointF CenterOfGravity(){if (_rleData.Count == 0) return PointF.Empty;double sumX = 0, sumY = 0;double totalPixels = 0;foreach (var kvp in _rleData){int y = kvp.Key;foreach (var interval in kvp.Value){int length = interval.Length;double centerX = (interval.Start + interval.End) / 2.0;sumX += centerX * length;sumY += y * length;totalPixels += length;}}if (totalPixels == 0) return PointF.Empty;return new PointF((float)(sumX / totalPixels), (float)(sumY / totalPixels));}// 计算圆度public double Circularity(){double area = Area();if (area == 0) return 0;// 计算周长 (简化版，实际需要更精确的计算)double perimeter = 0;foreach (var kvp in _rleData){int y = kvp.Key;foreach (var interval in kvp.Value){// 上边界if (!_rleData.ContainsKey(y - 1) || !HasPixel(y - 1, interval.Start, interval.End))perimeter += interval.Length;// 下边界if (!_rleData.ContainsKey(y + 1) || !HasPixel(y + 1, interval.Start, interval.End))perimeter += interval.Length;// 左边界perimeter += CountLeftEdges(interval.Start, y);// 右边界perimeter += CountRightEdges(interval.End, y);}}return (4 * Math.PI * area) / (perimeter * perimeter);}private bool HasPixel(int y, int startX, int endX){if (!_rleData.TryGetValue(y, out var intervals))return false;foreach (var interval in intervals){if (interval.Start <= endX && interval.End >= startX)return true;}return false;}private int CountLeftEdges(int x, int y){int count = 0;if (!HasPixel(y, x - 1, x - 1)) count++;return count;}private int CountRightEdges(int x, int y){int count = 0;if (!HasPixel(y, x + 1, x + 1)) count++;return count;}// 转换为位图public Bitmap ToBitmap(Bitmap bitmap, Color color ,bool clear=true){if (_rleData.Count == 0) return new Bitmap(1, 1);Rectangle bb = BoundingBox;//Bitmap bitmap = new Bitmap(bb.Width, bb.Height);if (clear){using (Graphics g = Graphics.FromImage(bitmap)){g.Clear(Color.Black);}}foreach (var kvp in _rleData){int y = kvp.Key ;foreach (var interval in kvp.Value){for (int x = interval.Start ; x <= interval.End ; x++){if (x >= 0 && x < bitmap.Width && y >= 0 && y < bitmap.Height){bitmap.SetPixel(x, y,color);}}}}return bitmap;}// 形态学操作 - 膨胀public HRegion Dilation(int radius){HRegion result = new HRegion();foreach (var kvp in _rleData){int y = kvp.Key;foreach (var interval in kvp.Value){for (int dy = -radius; dy <= radius; dy++){int newY = y + dy;int dx = (int)Math.Sqrt(radius * radius - dy * dy);int newStart = interval.Start - dx;int newEnd = interval.End + dx;if (!result._rleData.TryGetValue(newY, out var intervals)){intervals = new List<Interval>();result._rleData[newY] = intervals;}// 合并重叠或相邻的区间bool merged = false;for (int i = 0; i < intervals.Count; i++){if (newEnd + 1 >= intervals[i].Start && newStart - 1 <= intervals[i].End){// 合并区间int start = Math.Min(newStart, intervals[i].Start);int end = Math.Max(newEnd, intervals[i].End);intervals[i] = new Interval(start, end);merged = true;// 检查是否需要与后面的区间合并while (i + 1 < intervals.Count && intervals[i].End + 1 >= intervals[i + 1].Start){intervals[i] = new Interval(intervals[i].Start, Math.Max(intervals[i].End, intervals[i + 1].End));intervals.RemoveAt(i + 1);}break;}}if (!merged){intervals.Add(new Interval(newStart, newEnd));// 保持区间有序intervals.Sort((a, b) => a.Start.CompareTo(b.Start));}}}}result._boundingBoxValid = false;return result;}// 腐蚀操作入口（自动选择最优策略）public HRegion ErosionOptimized(int radius, bool useCircle = true){if (radius <= 0) return new HRegion();if (_rleData.Count == 0) return new HRegion();// 超大半径直接返回空（优化极端情况）Rectangle bb = BoundingBox;if (radius >= Math.Max(bb.Width, bb.Height) * 2)return new HRegion();// 分阶段腐蚀（自动分解大半径）HRegion result = this;int remainingRadius = radius;while (remainingRadius > 0){int currentRadius = Math.Min(5, remainingRadius); // 单次最大半径=5result = useCircle ?ErosionCircleSingleStep(result, currentRadius) :ErosionRectSingleStep(result, currentRadius);remainingRadius -= currentRadius;}return result;}// 单步圆形腐蚀（并行优化）private HRegion ErosionCircleSingleStep(HRegion region, int radius){var result = new ConcurrentDictionary<int, List<Interval>>();var offsets = PrecomputeCircleOffsets(radius); // 预计算圆形偏移量Parallel.ForEach(region._rleData.Keys, y =>{var erodedIntervals = new List<Interval>();foreach (var interval in region._rleData[y]){int currentStart = -1;for (int x = interval.Start; x <= interval.End; x++){bool isValid = true;foreach (var (dx, dy) in offsets){int checkX = x + dx;int checkY = y + dy;if (!region.IsPixelCovered(checkY, checkX)){isValid = false;break;}}// 合并连续区间if (isValid){if (currentStart == -1)currentStart = x;}else if (currentStart != -1){erodedIntervals.Add(new Interval(currentStart, x - 1));currentStart = -1;}}if (currentStart != -1)erodedIntervals.Add(new Interval(currentStart, interval.End));}if (erodedIntervals.Count > 0)result[y] = MergeIntervals(erodedIntervals);});return new HRegion { _rleData = new Dictionary<int, List<Interval>>(result) };}// 单步矩形腐蚀（分离为水平+垂直）private HRegion ErosionRectSingleStep(HRegion region, int radius){// 水平腐蚀（1x(2r+1)矩形）var horiEroded = ErosionRect1D(region, radius, 0);// 垂直腐蚀（(2r+1)x1矩形）var vertEroded = ErosionRect1D(horiEroded, 0, radius);return vertEroded;}// 一维矩形腐蚀（高效实现）private HRegion ErosionRect1D(HRegion region, int radiusX, int radiusY){var result = new ConcurrentDictionary<int, List<Interval>>();int totalRadius = radiusX + radiusY;Parallel.ForEach(region._rleData.Keys, y =>{var erodedIntervals = new List<Interval>();foreach (var interval in region._rleData[y]){int start = interval.Start + radiusX;int end = interval.End - radiusX;if (start > end) continue;// 检查垂直方向bool isValid = true;for (int dy = -radiusY; dy <= radiusY; dy++){int checkY = y + dy;if (!region.IsIntervalCovered(checkY, start, end)){isValid = false;break;}}if (isValid)erodedIntervals.Add(new Interval(start, end));}if (erodedIntervals.Count > 0)result[y] = MergeIntervals(erodedIntervals);});return new HRegion { _rleData = new Dictionary<int, List<Interval>>(result) };}// 预计算圆形结构元素偏移量private List<(int dx, int dy)> PrecomputeCircleOffsets(int radius){var offsets = new List<(int, int)>();int r2 = radius * radius;for (int dy = -radius; dy <= radius; dy++)for (int dx = -radius; dx <= radius; dx++)if (dx * dx + dy * dy <= r2)offsets.Add((dx, dy));return offsets;}// 合并重叠/相邻区间private List<Interval> MergeIntervals(List<Interval> intervals){if (intervals.Count == 0) return intervals;var merged = new List<Interval> { intervals[0] };for (int i = 1; i < intervals.Count; i++){var last = merged[merged.Count - 1];if (intervals[i].Start <= last.End + 1)merged[merged.Count - 1] = new Interval(last.Start, Math.Max(last.End, intervals[i].End));elsemerged.Add(intervals[i]);}return merged;}// 形态学操作 - 腐蚀// 形态学操作 - 腐蚀（支持圆形和矩形结构元素）public HRegion Erosion(int radius, bool useCircle = true){HRegion result = new HRegion();foreach (var kvp in _rleData){int y = kvp.Key;foreach (var interval in kvp.Value){// 初始候选区间（先应用半径的简单收缩）int candidateStart = interval.Start + radius;int candidateEnd = interval.End - radius;if (candidateStart > candidateEnd) continue;// 检查候选区间内的每个点List<Interval> validIntervals = new List<Interval>();int currentStart = -1;int currentEnd = -1;for (int x = candidateStart; x <= candidateEnd; x++){bool isValid = true;if (useCircle){// 圆形结构元素检查for (int dy = -radius; dy <= radius; dy++){for (int dx = -radius; dx <= radius; dx++){// 检查是否在圆形内if (dx * dx + dy * dy <= radius * radius){int checkX = x + dx;int checkY = y + dy;if (!HasPixel(checkY, checkX, checkX)){isValid = false;break;}}}if (!isValid) break;}}else{// 矩形结构元素检查for (int dy = -radius; dy <= radius; dy++){for (int dx = -radius; dx <= radius; dx++){int checkX = x + dx;int checkY = y + dy;if (!HasPixel(checkY, checkX, checkX)){isValid = false;break;}}if (!isValid) break;}}// 处理连续的有效区间if (isValid){if (currentStart == -1){currentStart = x;currentEnd = x;}else if (x == currentEnd + 1){currentEnd = x;}else{validIntervals.Add(new Interval(currentStart, currentEnd));currentStart = x;currentEnd = x;}}}// 添加最后一个区间if (currentStart != -1){validIntervals.Add(new Interval(currentStart, currentEnd));}// 将有效区间添加到结果中if (validIntervals.Count > 0){if (!result._rleData.TryGetValue(y, out var resultIntervals)){resultIntervals = new List<Interval>();result._rleData[y] = resultIntervals;}resultIntervals.AddRange(validIntervals);}}}// 对每行的区间进行合并和排序foreach (var intervals in result._rleData.Values){intervals.Sort((a, b) => a.Start.CompareTo(b.Start));// 合并重叠或相邻的区间for (int i = 0; i < intervals.Count - 1;){if (intervals[i].End + 1 >= intervals[i + 1].Start){intervals[i] = new Interval(Math.Min(intervals[i].Start, intervals[i + 1].Start),Math.Max(intervals[i].End, intervals[i + 1].End));intervals.RemoveAt(i + 1);}else{i++;}}}result._boundingBoxValid = false;return result;}/// <summary>/// 检查指定像素 (x,y) 是否被当前区域覆盖/// </summary>/// <param name="y">行坐标</param>/// <param name="x">列坐标</param>/// <returns>true如果像素在区域内</returns>public bool IsPixelCovered(int y, int x){// 1. 检查该行是否存在数据if (!_rleData.TryGetValue(y, out var intervals))return false;// 2. 二分查找优化：快速定位x可能所在的区间int left = 0;int right = intervals.Count - 1;while (left <= right){int mid = (left + right) / 2;var interval = intervals[mid];if (x < interval.Start){right = mid - 1;}else if (x > interval.End){left = mid + 1;}else{return true; // x在区间[Start, End]内}}return false;}/// <summary>/// 检查指定行y上的区间[start, end]是否完全被覆盖/// </summary>public bool IsIntervalCovered(int y, int start, int end){if (!_rleData.TryGetValue(y, out var intervals))return false;// 遍历所有区间，检查是否有一个区间完全包含[start, end]foreach (var interval in intervals){if (interval.Start <= start && interval.End >= end)return true;// 区间已排序，如果当前区间.Start > end，后续不可能有覆盖if (interval.Start > end)break;}return false;}// 开运算 = 先腐蚀后膨胀public HRegion Opening(int radius){return Erosion(radius).Dilation(radius);}// 闭运算 = 先膨胀后腐蚀public HRegion Closing(int radius){return Dilation(radius).Erosion(radius);}// 连通域分析public static List<HRegion> Connection(HRegion region){List<HRegion> connectedRegions = new List<HRegion>();if (region._rleData.Count == 0) return connectedRegions;// 使用并查集算法标记连通区域Dictionary<Point, int> labels = new Dictionary<Point, int>();int currentLabel = 1;UnionFind uf = new UnionFind();// 第一遍扫描: 分配临时标签foreach (var kvp in region._rleData){int y = kvp.Key;foreach (var interval in kvp.Value){for (int x = interval.Start; x <= interval.End; x++){Point p = new Point(x, y);// 检查相邻像素的标签List<int> neighborLabels = new List<int>();// 左邻if (labels.TryGetValue(new Point(x - 1, y), out int leftLabel))neighborLabels.Add(leftLabel);// 上邻if (labels.TryGetValue(new Point(x, y - 1), out int topLabel))neighborLabels.Add(topLabel);if (neighborLabels.Count == 0){// 新标签int newLabel = currentLabel++;labels[p] = newLabel;uf.MakeSet(newLabel);}else{// 使用最小的邻居标签int minLabel = neighborLabels.Min();labels[p] = minLabel;// 合并等价类foreach (var label in neighborLabels){if (label != minLabel)uf.Union(minLabel, label);}}}}}// 第二遍扫描: 分配最终标签并收集区域Dictionary<int, HRegion> labelToRegion = new Dictionary<int, HRegion>();foreach (var kvp in labels){Point p = kvp.Key;int label = uf.Find(kvp.Value);if (!labelToRegion.TryGetValue(label, out HRegion connectedRegion)){connectedRegion = new HRegion();labelToRegion[label] = connectedRegion;}if (!connectedRegion._rleData.TryGetValue(p.Y, out var intervals)){intervals = new List<Interval>();connectedRegion._rleData[p.Y] = intervals;}// 尝试合并相邻的像素bool merged = false;for (int i = 0; i < intervals.Count; i++){if (intervals[i].End + 1 == p.X){intervals[i] = new Interval(intervals[i].Start, p.X);merged = true;break;}else if (intervals[i].Start - 1 == p.X){intervals[i] = new Interval(p.X, intervals[i].End);merged = true;break;}}if (!merged){intervals.Add(new Interval(p.X, p.X));// 保持区间有序intervals.Sort((a, b) => a.Start.CompareTo(b.Start));}}// 更新每个连通区域的边界框foreach (var r in labelToRegion.Values){r.UpdateBoundingBox();connectedRegions.Add(r);}return connectedRegions;}// 并查集数据结构用于连通域分析private class UnionFind{private Dictionary<int, int> parent = new Dictionary<int, int>();private Dictionary<int, int> rank = new Dictionary<int, int>();public void MakeSet(int x){parent[x] = x;rank[x] = 0;}public int Find(int x){if (parent[x] != x)parent[x] = Find(parent[x]);return parent[x];}public void Union(int x, int y){int xRoot = Find(x);int yRoot = Find(y);if (xRoot == yRoot) return;if (rank[xRoot] < rank[yRoot])parent[xRoot] = yRoot;else if (rank[xRoot] > rank[yRoot])parent[yRoot] = xRoot;else{parent[yRoot] = xRoot;rank[xRoot]++;}}}}
}

测试代码

运行速度在300ms左右，有点儿慢

using System;
using System.Collections.Generic;
using System.ComponentModel;
using System.Data;
using System.Drawing;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using System.Windows.Forms;
using static System.Net.Mime.MediaTypeNames;
using Image = System.Drawing.Image;
using System.Diagnostics; // 添加Stopwatch命名空间namespace view3d
{public partial class MainForm : Form{HRegion region = new HRegion();Bitmap image;public MainForm(){InitializeComponent();}private void load_Click(object sender, EventArgs e){var stopwatch = Stopwatch.StartNew(); // 开始计时// 创建打开文件对话框OpenFileDialog openFileDialog = new OpenFileDialog();// 设置文件过滤器（只显示图片文件）openFileDialog.Filter = "Image Files|*.jpg;*.jpeg;*.png;*.bmp;*.gif";// 设置对话框标题openFileDialog.Title = "选择要加载的图片";// 显示对话框并检查用户是否点击了确定if (openFileDialog.ShowDialog() == DialogResult.OK){try{Image loadedImage = Image.FromFile(openFileDialog.FileName);// 从文件路径创建Image对象image = new Bitmap(loadedImage);string v1 = textBox1.Text;string v2 = textBox2.Text;byte i1 = byte.Parse(v1);byte i2 = byte.Parse(v2);region.FromBitmap((Bitmap)image, i1, i2);// 将图片显示到pictureBox1pictureBox1.Image = region.ToBitmap((Bitmap)image, Color.White);// 设置PictureBox的SizeMode为Zoom，使图片适应控件大小pictureBox1.SizeMode = PictureBoxSizeMode.Zoom;stopwatch.Stop(); // 停止计时MessageBox.Show($"图片加载和处理耗时: {stopwatch.ElapsedMilliseconds} 毫秒", "耗时统计");}catch (Exception ex){// 如果加载失败，显示错误信息MessageBox.Show("加载图片时出错: " + ex.Message, "错误",MessageBoxButtons.OK, MessageBoxIcon.Error);}}}private void pictureBox1_Click(object sender, EventArgs e){}private void connection_Click(object sender, EventArgs e){var stopwatch = Stopwatch.StartNew(); // 开始计时List<HRegion> conns = HRegion.Connection(region);using (Graphics g = Graphics.FromImage(image)){g.Clear(Color.Black);}// 预定义一组颜色，或者随机生成颜色Color[] colors = new Color[]{Color.Red,Color.Green,Color.Blue,Color.Yellow,Color.Magenta,Color.Cyan,Color.Orange,Color.Purple,Color.Lime,Color.Pink};for (int i = 0; i < conns.Count; i++){HRegion region = conns[i];if (region != null){// 使用循环索引选择颜色，如果超过预定义颜色数量则随机生成Color regionColor = i < colors.Length ? colors[i] : GetRandomColor();region.ToBitmap(image, regionColor, false);}}stopwatch.Stop(); // 停止计时MessageBox.Show($"连通域分析耗时: {stopwatch.ElapsedMilliseconds} 毫秒", "耗时统计");// 刷新显示pictureBox1.Invalidate();}// 辅助方法：生成随机颜色private Color GetRandomColor(){Random rand = new Random();return Color.FromArgb(rand.Next(256), rand.Next(256), rand.Next(256));}private void Dilation_Click(object sender, EventArgs e){var stopwatch = Stopwatch.StartNew(); // 开始计时string v1 = textBox1.Text;string v2 = textBox2.Text;byte i1 = byte.Parse(v1);byte i2 = byte.Parse(v2);region = region.Dilation(i2);region.ToBitmap(image, Color.Yellow);stopwatch.Stop(); // 停止计时MessageBox.Show($"膨胀操作耗时: {stopwatch.ElapsedMilliseconds} 毫秒", "耗时统计");// 刷新显示pictureBox1.Invalidate();}private void button2_Click(object sender, EventArgs e){var stopwatch = Stopwatch.StartNew(); // 开始计时string v1 = textBox1.Text;string v2 = textBox2.Text;byte i1 = byte.Parse(v1);byte i2 = byte.Parse(v2);region = region.ErosionOptimized(i2);region.ToBitmap(image, Color.Green);stopwatch.Stop(); // 停止计时MessageBox.Show($"腐蚀操作耗时: {stopwatch.ElapsedMilliseconds} 毫秒", "耗时统计");// 刷新显示pictureBox1.Invalidate();}}
}

测试demo下载

见资源标定