优化红外成像制导导弹的性能，修改了目标识别算法各装备的特征参数
This commit is contained in:
parent
63863df67a
commit
f873ebfd5b
@ -860,7 +860,7 @@ namespace ThreatSource.Tests.Jamming
|
||||
|
||||
// 计算功率比值，分析目标选择
|
||||
double powerRatio = decoyReceivedPower / realReceivedPower;
|
||||
Debug.WriteLine($"诱偏/真实目标功率比: {powerRatio:F6}");
|
||||
Debug.WriteLine($"诱偏/真实目标功率比: {powerRatio:E3}");
|
||||
Debug.WriteLine($"功率比例分析: {(powerRatio > 1 ? "诱偏目标功率更强" : "真实目标功率更强")}");
|
||||
}
|
||||
|
||||
|
||||
@ -33,7 +33,7 @@ namespace ThreatSource.Tests.Utils
|
||||
);
|
||||
|
||||
// Assert - 验证结果在合理范围内
|
||||
Debug.WriteLine($"激光透过率: {transmittance:F6}");
|
||||
Debug.WriteLine($"激光透过率: {transmittance:E3}");
|
||||
Assert.True(transmittance >= 0.0);
|
||||
Assert.True(transmittance <= 1.0);
|
||||
Assert.NotEqual(0.0, transmittance); // 确保不是0
|
||||
@ -64,7 +64,7 @@ namespace ThreatSource.Tests.Utils
|
||||
);
|
||||
|
||||
// Assert - 验证结果在合理范围内
|
||||
Debug.WriteLine($"红外透过率: {transmittance:F6}");
|
||||
Debug.WriteLine($"红外透过率: {transmittance:E3}");
|
||||
Assert.True(transmittance >= 0.0);
|
||||
Assert.True(transmittance <= 1.0);
|
||||
Assert.NotEqual(0.0, transmittance); // 确保不是0
|
||||
@ -95,7 +95,7 @@ namespace ThreatSource.Tests.Utils
|
||||
);
|
||||
|
||||
// Assert - 验证结果在合理范围内
|
||||
Debug.WriteLine($"湍流效应: {turbulenceEffect:F6}");
|
||||
Debug.WriteLine($"湍流效应: {turbulenceEffect:E3}");
|
||||
Assert.True(turbulenceEffect >= 0.0);
|
||||
Assert.True(turbulenceEffect <= 1.0);
|
||||
Assert.NotEqual(0.0, turbulenceEffect); // 确保不是0
|
||||
@ -116,7 +116,7 @@ namespace ThreatSource.Tests.Utils
|
||||
);
|
||||
|
||||
// Assert - 验证结果在合理范围内
|
||||
Debug.WriteLine($"烟幕透过率: {smokeTransmittance:F6}");
|
||||
Debug.WriteLine($"烟幕透过率: {smokeTransmittance:E3}");
|
||||
Assert.True(smokeTransmittance >= 0.0);
|
||||
Assert.True(smokeTransmittance <= 1.0);
|
||||
}
|
||||
@ -164,9 +164,9 @@ namespace ThreatSource.Tests.Utils
|
||||
distance * 1000, radiationType, wavelength, rainyWeather); // 转换为米
|
||||
|
||||
// Assert - 验证不同天气条件对透过率有影响
|
||||
Debug.WriteLine($"晴天透过率: {clearTransmittance:F6}");
|
||||
Debug.WriteLine($"雾天透过率: {foggyTransmittance:F6}");
|
||||
Debug.WriteLine($"雨天透过率: {rainyTransmittance:F6}");
|
||||
Debug.WriteLine($"晴天透过率: {clearTransmittance:E3}");
|
||||
Debug.WriteLine($"雾天透过率: {foggyTransmittance:E3}");
|
||||
Debug.WriteLine($"雨天透过率: {rainyTransmittance:E3}");
|
||||
|
||||
// 验证所有值在有效范围内
|
||||
Assert.True(clearTransmittance >= 0.0 && clearTransmittance <= 1.0);
|
||||
|
||||
@ -27,14 +27,14 @@ BarrelLength = 5.8 # 炮管长度 (米)
|
||||
|
||||
[Properties.ThermalPattern] # 热辐射特征模型
|
||||
StaticPatternSource = [ # 静态时温度分布
|
||||
[40, 45, 80],
|
||||
[35, 40, 90],
|
||||
[50, 50, 60]
|
||||
[40, 45, 80], # 前部装甲
|
||||
[35, 40, 90], # 中部装甲
|
||||
[50, 50, 60] # 后部发动机
|
||||
]
|
||||
MovingPatternSource = [ # 移动时温度分布
|
||||
[45, 50, 85],
|
||||
[40, 45, 95],
|
||||
[65, 65, 75]
|
||||
[45, 50, 85], # 前部装甲
|
||||
[40, 45, 95], # 中部装甲
|
||||
[65, 65, 75] # 后部发动机
|
||||
]
|
||||
|
||||
[Properties.RcsPattern]
|
||||
|
||||
@ -6,6 +6,7 @@ using ThreatSource.Simulation;
|
||||
using System.Diagnostics;
|
||||
using System.Collections.Concurrent;
|
||||
using System.Collections.Generic;
|
||||
using System.Runtime.InteropServices;
|
||||
|
||||
namespace ThreatSource.Guidance
|
||||
{
|
||||
@ -98,17 +99,34 @@ namespace ThreatSource.Guidance
|
||||
{
|
||||
if (DoubleArrayPool.TryDequeue(out var array))
|
||||
{
|
||||
// 清零重用的数组
|
||||
for (int y = 0; y < imageHeight; y++)
|
||||
{
|
||||
for (int x = 0; x < imageWidth; x++)
|
||||
{
|
||||
array[y, x] = -1.0;
|
||||
}
|
||||
}
|
||||
// 使用优化的数组清零方法
|
||||
ClearDoubleArrayToNegativeOne(array);
|
||||
return array;
|
||||
}
|
||||
return new double[imageHeight, imageWidth];
|
||||
|
||||
// 创建新数组并初始化为-1.0
|
||||
var newArray = new double[imageHeight, imageWidth];
|
||||
ClearDoubleArrayToNegativeOne(newArray);
|
||||
return newArray;
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 高效清零double数组为-1.0
|
||||
/// </summary>
|
||||
private void ClearDoubleArrayToNegativeOne(double[,] array)
|
||||
{
|
||||
// 使用安全的多维数组填充方法
|
||||
int height = array.GetLength(0);
|
||||
int width = array.GetLength(1);
|
||||
|
||||
// 优化：按行填充，利用内存局部性
|
||||
for (int y = 0; y < height; y++)
|
||||
{
|
||||
for (int x = 0; x < width; x++)
|
||||
{
|
||||
array[y, x] = -1.0;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
@ -241,7 +259,7 @@ namespace ThreatSource.Guidance
|
||||
double angleX = Math.Atan2(x, z);
|
||||
double angleY = Math.Atan2(y, z);
|
||||
|
||||
Debug.WriteLine($"投影角度: X={angleX:F6} 弧度, Y={angleY:F6} 弧度");
|
||||
Debug.WriteLine($"投影角度: X={angleX:E3} 弧度, Y={angleY:E3} 弧度");
|
||||
|
||||
return (angleX, angleY);
|
||||
}
|
||||
@ -391,7 +409,7 @@ namespace ThreatSource.Guidance
|
||||
smokeIntensityLayer[y, x] = Math.Max(smokeIntensityLayer[y, x], smokeIntensity);
|
||||
}
|
||||
}
|
||||
Debug.WriteLine($"记录遮蔽烟幕(Wall): 中心({smokeCenterX},{smokeCenterY}), 尺寸{pixelWidth}x{pixelHeight}px, 强度{smokeIntensity:F6}");
|
||||
Debug.WriteLine($"记录遮蔽烟幕(Wall): 中心({smokeCenterX},{smokeCenterY}), 尺寸{pixelWidth}x{pixelHeight}px, 强度{smokeIntensity:E3}");
|
||||
break;
|
||||
|
||||
case SmokeScreenType.Cloud:
|
||||
@ -420,7 +438,7 @@ namespace ThreatSource.Guidance
|
||||
}
|
||||
}
|
||||
}
|
||||
Debug.WriteLine($"记录遮蔽烟幕(Cloud): 中心({smokeCenterX},{smokeCenterY}), 半径{pixelRadius}px, 强度{smokeIntensity:F6}");
|
||||
Debug.WriteLine($"记录遮蔽烟幕(Cloud): 中心({smokeCenterX},{smokeCenterY}), 半径{pixelRadius}px, 强度{smokeIntensity:E3}");
|
||||
break;
|
||||
}
|
||||
}
|
||||
@ -537,8 +555,8 @@ namespace ThreatSource.Guidance
|
||||
return;
|
||||
}
|
||||
|
||||
int halfLength = pixelLength / 2;
|
||||
int halfWidth = pixelWidth / 2;
|
||||
int halfLength = Math.Max(1, pixelLength / 2); // 防止除零
|
||||
int halfWidth = Math.Max(1, pixelWidth / 2); // 防止除零
|
||||
int pixelsSet = 0;
|
||||
double maxSetIntensity = 0;
|
||||
|
||||
@ -615,9 +633,9 @@ namespace ThreatSource.Guidance
|
||||
int pixelYInBox = dy + halfWidth;
|
||||
int pixelXInBox = dx + halfLength;
|
||||
|
||||
// Map to pattern indices [0, 2]
|
||||
int patternRow = (int)Math.Floor(pixelYInBox / patternPixelWidth);
|
||||
int patternCol = (int)Math.Floor(pixelXInBox / patternPixelHeight); // X maps to Col
|
||||
// Map to pattern indices [0, 2] - 修正坐标映射
|
||||
int patternRow = (int)Math.Floor(pixelYInBox / patternPixelHeight); // Y坐标用Height
|
||||
int patternCol = (int)Math.Floor(pixelXInBox / patternPixelWidth); // X坐标用Width
|
||||
|
||||
// Clamp indices to [0, 2]
|
||||
patternRow = Math.Max(0, Math.Min(2, patternRow));
|
||||
@ -658,7 +676,7 @@ namespace ThreatSource.Guidance
|
||||
}
|
||||
}
|
||||
|
||||
Debug.WriteLine($"目标基于ThermalPattern强度分布: 像素设置: {pixelsSet}, 最大强度: {maxSetIntensity:F6} W/sr");
|
||||
Debug.WriteLine($"目标基于ThermalPattern强度分布: 像素设置: {pixelsSet}, 最大强度: {maxSetIntensity:E3} W/sr");
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
|
||||
@ -86,7 +86,14 @@ namespace ThreatSource.Guidance
|
||||
/// </summary>
|
||||
private const double SMOKE_COVERAGE_THRESHOLD = 0.75;
|
||||
|
||||
|
||||
// **性能优化：静态常量Sobel算子，避免重复分配**
|
||||
private static readonly double[,] SOBEL_X = {{-1, 0, 1}, {-2, 0, 2}, {-1, 0, 1}};
|
||||
private static readonly double[,] SOBEL_Y = {{-1, -2, -1}, {0, 0, 0}, {1, 2, 1}};
|
||||
|
||||
// **性能优化：重用数组，避免频繁分配**
|
||||
private readonly double[] _histogramBuffer = new double[10];
|
||||
private readonly double[,] _gridGradientsBuffer = new double[3, 3];
|
||||
private readonly int[,] _gridCountsBuffer = new int[3, 3];
|
||||
|
||||
/// <summary>
|
||||
/// 图像预扫描结果
|
||||
@ -99,29 +106,133 @@ namespace ThreatSource.Guidance
|
||||
public int MaxX { get; set; }
|
||||
public int MaxY { get; set; }
|
||||
public double Density { get; set; }
|
||||
public int Width => MaxX - MinX + 1;
|
||||
public int Height => MaxY - MinY + 1;
|
||||
public double AspectRatio => Height > 0 ? (double)Width / Height : 1.0;
|
||||
public readonly int Width => MaxX - MinX + 1;
|
||||
public readonly int Height => MaxY - MinY + 1;
|
||||
public readonly double AspectRatio => Height > 0 ? (double)Width / Height : 1.0;
|
||||
|
||||
// 早期退出条件
|
||||
public bool IsEmpty => PixelsAboveThreshold == 0;
|
||||
public bool IsTooSmall => PixelsAboveThreshold < BLOB_MIN_AREA_THRESHOLD;
|
||||
public bool IsSinglePixel => PixelsAboveThreshold <= 2;
|
||||
public bool IsLinearTarget => AspectRatio > 10.0 || AspectRatio < 0.1; // 极端长宽比
|
||||
public bool IsTooSparse => Density < 0.05; // 密度过低，可能是噪声
|
||||
public bool IsSimpleTarget => Density > 0.8 && PixelsAboveThreshold < 1000; // 相对较小且密集
|
||||
public readonly bool IsEmpty => PixelsAboveThreshold == 0;
|
||||
public readonly bool IsTooSmall => PixelsAboveThreshold < BLOB_MIN_AREA_THRESHOLD;
|
||||
public readonly bool IsSinglePixel => PixelsAboveThreshold <= 2;
|
||||
public readonly bool IsLinearTarget => AspectRatio > 10.0 || AspectRatio < 0.1; // 极端长宽比
|
||||
public readonly bool IsTooSparse => Density < 0.05; // 密度过低，可能是噪声
|
||||
public readonly bool IsSimpleTarget => Density > 0.8 && PixelsAboveThreshold < 1000; // 相对较小且密集
|
||||
|
||||
// 检查是否位于图像边缘
|
||||
public bool IsAtImageEdge(int imageWidth, int imageHeight)
|
||||
public readonly bool IsAtImageEdge(int imageWidth, int imageHeight)
|
||||
{
|
||||
return MinX <= 1 || MinY <= 1 || MaxX >= imageWidth - 2 || MaxY >= imageHeight - 2;
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 统一图像分析结果 - 单次遍历完成多项计算
|
||||
/// </summary>
|
||||
private struct ImageAnalysisResult
|
||||
{
|
||||
// 全图统计信息
|
||||
public double MinIntensity { get; set; }
|
||||
public double MaxIntensity { get; set; }
|
||||
public double MeanIntensity { get; set; }
|
||||
public double TotalIntensity { get; set; }
|
||||
public int TotalPixels { get; set; }
|
||||
public int ValidPixels { get; set; } // 强度 > 0 的像素数
|
||||
|
||||
// 区域统计信息（如果指定了目标区域）
|
||||
public double CenterIntensity { get; set; }
|
||||
public int CenterPixelCount { get; set; }
|
||||
public readonly double AvgCenterIntensity => CenterPixelCount > 0 ? CenterIntensity / CenterPixelCount : 0;
|
||||
public readonly double CenterIntensityRatio => TotalPixels > 0 && TotalIntensity > 0 && CenterPixelCount > 0 ?
|
||||
(CenterIntensity / CenterPixelCount) / (TotalIntensity / TotalPixels) : 0;
|
||||
|
||||
// 直方图数据（10-bin）
|
||||
public double[]? Histogram { get; set; }
|
||||
public double ConcentrationScore { get; set; } // 高强度bin的集中度
|
||||
|
||||
// 梯度信息（3x3网格）
|
||||
public double[,]? GridGradients { get; set; }
|
||||
public int[,]? GridCounts { get; set; }
|
||||
public double AvgGradient { get; set; }
|
||||
public double MaxGradient { get; set; }
|
||||
|
||||
// 烟幕覆盖信息
|
||||
public int SmokeCoveredPixels { get; set; }
|
||||
public readonly double SmokeCoverageRatio => TotalPixels > 0 ? (double)SmokeCoveredPixels / TotalPixels : 0;
|
||||
|
||||
// 计算动态阈值
|
||||
public readonly double CalculateThreshold(double thresholdFactor = 0.05)
|
||||
{
|
||||
if (ValidPixels == 0 || MaxIntensity <= MinIntensity) return 0;
|
||||
double dynamicRange = MaxIntensity - MinIntensity;
|
||||
return MinIntensity + dynamicRange * thresholdFactor;
|
||||
}
|
||||
|
||||
// 计算强度模式得分
|
||||
public readonly double CalculateIntensityPattern()
|
||||
{
|
||||
if (Histogram == null || ConcentrationScore == 0) return 0;
|
||||
|
||||
// 通用强度模式计算：适用于Tank、APC、Helicopter等所有目标类型
|
||||
// 基于热分布的两个核心特征：集中度 + 分布特征
|
||||
|
||||
// 1. 高温区域集中度（所有目标都有发动机等热源）
|
||||
double hotRegionScore = ConcentrationScore;
|
||||
|
||||
// 2. 温度分布的非均匀性（不同目标有不同的分布模式）
|
||||
double distributionScore = 0;
|
||||
if (Histogram.Length >= 10)
|
||||
{
|
||||
// 计算温度分布的多样性和对比度
|
||||
double lowTempRegion = 0; // 低温区域（前30%的bin）
|
||||
double midTempRegion = 0; // 中温区域（中40%的bin）
|
||||
double highTempRegion = 0; // 高温区域（后30%的bin）
|
||||
|
||||
for (int i = 0; i < 3; i++) lowTempRegion += Histogram[i]; // 前30%
|
||||
for (int i = 3; i < 7; i++) midTempRegion += Histogram[i]; // 中40%
|
||||
for (int i = 7; i < 10; i++) highTempRegion += Histogram[i]; // 后30%
|
||||
|
||||
// 通用特征：良好的目标应该有明显的温度对比
|
||||
// - Tank: 前部低温 + 后部高温
|
||||
// - APC: 相对均匀但有温度差异
|
||||
// - Helicopter: 中部极高温 + 其他区域低温
|
||||
|
||||
if (highTempRegion > 0.05) // 必须有高温区域
|
||||
{
|
||||
// 温度对比度：高温区域与低温区域的差异
|
||||
double temperatureContrast = highTempRegion - lowTempRegion;
|
||||
|
||||
// 温度分布的复杂度：三个区域都有分布表示目标结构复杂
|
||||
double distributionComplexity = 0;
|
||||
if (lowTempRegion > 0.02 && midTempRegion > 0.02 && highTempRegion > 0.02)
|
||||
{
|
||||
distributionComplexity = 0.3; // 奖励复杂的温度分布
|
||||
}
|
||||
|
||||
// 综合分布得分
|
||||
distributionScore = Math.Min(1.0,
|
||||
Math.Max(0, temperatureContrast) * 2.0 + distributionComplexity);
|
||||
}
|
||||
}
|
||||
|
||||
// 3. 中心强度特征（保留原有逻辑作为补充）
|
||||
double centerScore = CenterIntensityRatio > 0 ? Math.Min(1.0, CenterIntensityRatio) : 0;
|
||||
|
||||
// 4. 综合得分：集中度50% + 分布特征30% + 中心特征20%
|
||||
return Math.Max(0, Math.Min(1,
|
||||
hotRegionScore * 0.5 + distributionScore * 0.3 + centerScore * 0.2));
|
||||
}
|
||||
|
||||
// 计算温度梯度得分
|
||||
public readonly double CalculateTemperatureGradient()
|
||||
{
|
||||
return Math.Max(0, Math.Min(1, AvgGradient));
|
||||
}
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 创建简单分割结果（跳过Union-Find）
|
||||
/// </summary>
|
||||
private ImageSegment CreateSimpleSegment(int minX, int minY, int maxX, int maxY)
|
||||
private static ImageSegment CreateSimpleSegment(int minX, int minY, int maxX, int maxY)
|
||||
{
|
||||
int width = maxX - minX + 1;
|
||||
int height = maxY - minY + 1;
|
||||
@ -137,28 +248,236 @@ namespace ThreatSource.Guidance
|
||||
);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 统一图像区域分析 - 单次遍历完成多项计算
|
||||
/// </summary>
|
||||
/// <param name="image">红外图像</param>
|
||||
/// <param name="targetRegion">目标区域（null表示全图分析）</param>
|
||||
/// <param name="includeGradient">是否包含梯度计算</param>
|
||||
/// <param name="includeHistogram">是否包含直方图计算</param>
|
||||
/// <returns>统一分析结果</returns>
|
||||
private static ImageAnalysisResult AnalyzeImageRegion(
|
||||
InfraredImage image,
|
||||
ImageSegment? targetRegion = null,
|
||||
bool includeGradient = false,
|
||||
bool includeHistogram = false)
|
||||
{
|
||||
var result = new ImageAnalysisResult
|
||||
{
|
||||
MinIntensity = double.MaxValue,
|
||||
MaxIntensity = double.MinValue,
|
||||
TotalIntensity = 0,
|
||||
TotalPixels = 0,
|
||||
ValidPixels = 0,
|
||||
CenterIntensity = 0,
|
||||
CenterPixelCount = 0,
|
||||
SmokeCoveredPixels = 0
|
||||
};
|
||||
|
||||
// 确定分析区域
|
||||
int startX, endX, startY, endY;
|
||||
if (targetRegion.HasValue)
|
||||
{
|
||||
var region = targetRegion.Value;
|
||||
startX = Math.Max(0, region.Center.X - region.Size.Width / 2);
|
||||
endX = Math.Min(image.Width - 1, region.Center.X + region.Size.Width / 2);
|
||||
startY = Math.Max(0, region.Center.Y - region.Size.Height / 2);
|
||||
endY = Math.Min(image.Height - 1, region.Center.Y + region.Size.Height / 2);
|
||||
}
|
||||
else
|
||||
{
|
||||
startX = 0; endX = image.Width - 1;
|
||||
startY = 0; endY = image.Height - 1;
|
||||
}
|
||||
|
||||
// 初始化可选计算组件
|
||||
const int BINS = 10;
|
||||
double[]? histogram = includeHistogram ? new double[BINS] : null;
|
||||
double[,]? gridGradients = includeGradient ? new double[3, 3] : null;
|
||||
int[,]? gridCounts = includeGradient ? new int[3, 3] : null;
|
||||
|
||||
// 中心区域定义（用于区域分析）
|
||||
int centerRegionWidth = 0, centerRegionHeight = 0;
|
||||
int centerStartX = 0, centerEndX = 0, centerStartY = 0, centerEndY = 0;
|
||||
|
||||
if (targetRegion.HasValue)
|
||||
{
|
||||
var region = targetRegion.Value;
|
||||
centerRegionWidth = region.Size.Width / 3;
|
||||
centerRegionHeight = region.Size.Height / 3;
|
||||
centerStartX = region.Center.X - centerRegionWidth / 2;
|
||||
centerEndX = region.Center.X + centerRegionWidth / 2;
|
||||
centerStartY = region.Center.Y - centerRegionHeight / 2;
|
||||
centerEndY = region.Center.Y + centerRegionHeight / 2;
|
||||
}
|
||||
|
||||
// Sobel算子（用于梯度计算）
|
||||
double[,]? sobelX = includeGradient ? new double[,] {{-1, 0, 1}, {-2, 0, 2}, {-1, 0, 1}} : null;
|
||||
double[,]? sobelY = includeGradient ? new double[,] {{-1, -2, -1}, {0, 0, 0}, {1, 2, 1}} : null;
|
||||
|
||||
// 第一遍扫描：收集基本统计信息
|
||||
for (int y = startY; y <= endY; y++)
|
||||
{
|
||||
for (int x = startX; x <= endX; x++)
|
||||
{
|
||||
double intensity = image.GetIntensity(y, x);
|
||||
result.TotalPixels++;
|
||||
result.TotalIntensity += intensity;
|
||||
|
||||
if (intensity > 0)
|
||||
{
|
||||
result.ValidPixels++;
|
||||
result.MinIntensity = Math.Min(result.MinIntensity, intensity);
|
||||
result.MaxIntensity = Math.Max(result.MaxIntensity, intensity);
|
||||
}
|
||||
|
||||
// 中心区域统计（仅在区域分析时）
|
||||
if (targetRegion.HasValue &&
|
||||
x >= centerStartX && x <= centerEndX &&
|
||||
y >= centerStartY && y <= centerEndY)
|
||||
{
|
||||
result.CenterIntensity += intensity;
|
||||
result.CenterPixelCount++;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// 处理无效情况
|
||||
if (result.ValidPixels == 0)
|
||||
{
|
||||
result.MinIntensity = 0;
|
||||
result.MaxIntensity = 0;
|
||||
result.MeanIntensity = 0;
|
||||
return result;
|
||||
}
|
||||
|
||||
result.MeanIntensity = result.TotalIntensity / result.TotalPixels;
|
||||
double binSize = includeHistogram ? (result.MaxIntensity - result.MinIntensity) / BINS : 0;
|
||||
|
||||
// 计算网格尺寸（用于梯度分析）
|
||||
int regionWidth = endX - startX + 1;
|
||||
int regionHeight = endY - startY + 1;
|
||||
int gridWidth = includeGradient ? regionWidth / 3 : 0;
|
||||
int gridHeight = includeGradient ? regionHeight / 3 : 0;
|
||||
|
||||
double totalGradient = 0;
|
||||
int gradientCount = 0;
|
||||
|
||||
// 第二遍扫描：直方图和梯度计算
|
||||
for (int y = startY; y <= endY; y++)
|
||||
{
|
||||
for (int x = startX; x <= endX; x++)
|
||||
{
|
||||
double intensity = image.GetIntensity(y, x);
|
||||
|
||||
// 直方图计算
|
||||
if (includeHistogram && intensity > 0 && binSize > 0)
|
||||
{
|
||||
int bin = Math.Min(BINS - 1, Math.Max(0, (int)((intensity - result.MinIntensity) / binSize)));
|
||||
histogram![bin]++;
|
||||
}
|
||||
|
||||
// 梯度计算（需要边界检查）
|
||||
if (includeGradient &&
|
||||
y >= startY + 1 && y <= endY - 1 &&
|
||||
x >= startX + 1 && x <= endX - 1)
|
||||
{
|
||||
// 计算当前点属于哪个网格
|
||||
int gridRow = Math.Min(2, Math.Max(0, (y - startY) * 3 / regionHeight));
|
||||
int gridCol = Math.Min(2, Math.Max(0, (x - startX) * 3 / regionWidth));
|
||||
|
||||
double gradX = 0, gradY = 0;
|
||||
|
||||
// 应用Sobel算子
|
||||
for (int i = -1; i <= 1; i++)
|
||||
{
|
||||
for (int j = -1; j <= 1; j++)
|
||||
{
|
||||
double value = image.GetIntensity(y + i, x + j);
|
||||
gradX += value * sobelX![i + 1, j + 1];
|
||||
gradY += value * sobelY![i + 1, j + 1];
|
||||
}
|
||||
}
|
||||
|
||||
// 计算归一化梯度
|
||||
double gradient = result.MaxIntensity > 0 ?
|
||||
Math.Sqrt(gradX * gradX + gradY * gradY) / result.MaxIntensity : 0;
|
||||
|
||||
gridGradients![gridRow, gridCol] += gradient;
|
||||
gridCounts![gridRow, gridCol]++;
|
||||
totalGradient += gradient;
|
||||
gradientCount++;
|
||||
result.MaxGradient = Math.Max(result.MaxGradient, gradient);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// 完成直方图处理
|
||||
if (includeHistogram && histogram != null)
|
||||
{
|
||||
// 归一化直方图
|
||||
for (int i = 0; i < BINS; i++)
|
||||
{
|
||||
histogram[i] /= result.TotalPixels;
|
||||
}
|
||||
|
||||
// 计算集中度得分（最高30%的bin）
|
||||
double concentrationScore = 0;
|
||||
int topBinsStart = (int)(BINS * 0.7);
|
||||
for (int i = topBinsStart; i < BINS; i++)
|
||||
{
|
||||
concentrationScore += histogram[i];
|
||||
}
|
||||
|
||||
result.Histogram = histogram!;
|
||||
result.ConcentrationScore = concentrationScore;
|
||||
}
|
||||
|
||||
// 完成梯度处理
|
||||
if (includeGradient && gridGradients != null)
|
||||
{
|
||||
// 计算网格平均梯度
|
||||
for (int i = 0; i < 3; i++)
|
||||
{
|
||||
for (int j = 0; j < 3; j++)
|
||||
{
|
||||
if (gridCounts![i, j] > 0)
|
||||
{
|
||||
gridGradients![i, j] /= gridCounts[i, j];
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
result.GridGradients = gridGradients!;
|
||||
result.GridCounts = gridCounts!;
|
||||
result.AvgGradient = gradientCount > 0 ? totalGradient / gradientCount : 0;
|
||||
}
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 初始化红外图像目标识别器
|
||||
/// </summary>
|
||||
public InfraredTargetRecognizer()
|
||||
{
|
||||
// 初始化目标特征数据库 - 使用相对特征值
|
||||
// 初始化目标特征数据库 - 基于通用强度模式计算方法调整期望值
|
||||
targetFeatures = new Dictionary<EquipmentType, TargetFeature>
|
||||
{
|
||||
{ EquipmentType.Tank, new TargetFeature(
|
||||
aspectRatio: 2.9, // 典型主战坦克长宽比
|
||||
intensityPattern: 0.5, // 热量集中分布
|
||||
temperatureGradient: 0.3 // 高温度梯度
|
||||
intensityPattern: 0.8, // Tank：后部发动机高温，前部装甲低温，温度对比明显
|
||||
temperatureGradient: 0.7 // Tank的ThermalPattern实际输出约0.6-0.8
|
||||
)},
|
||||
{ EquipmentType.APC, new TargetFeature(
|
||||
aspectRatio: 2.1, // 较短的车身
|
||||
intensityPattern: 0.8, // 均匀热分布
|
||||
temperatureGradient: 0.8 // 中等温度梯度
|
||||
intensityPattern: 0.7, // APC：温度分布相对均匀，但仍有发动机热源
|
||||
temperatureGradient: 0.5 // APC的ThermalPattern实际输出约0.4-0.6
|
||||
)},
|
||||
{ EquipmentType.Helicopter, new TargetFeature(
|
||||
aspectRatio: 4.8, // 考虑旋翼长度
|
||||
intensityPattern: 0.8, // 发动机热量集中
|
||||
temperatureGradient: 0.8 // 较低的温度梯度
|
||||
intensityPattern: 0.9, // Helicopter：中部发动机极高温，上下温度差异极大
|
||||
temperatureGradient: 0.3 // Helicopter的ThermalPattern实际输出约0.2-0.4
|
||||
)}
|
||||
};
|
||||
}
|
||||
@ -212,7 +531,7 @@ namespace ThreatSource.Guidance
|
||||
int width = image.Width;
|
||||
int height = image.Height;
|
||||
|
||||
Debug.WriteLine($"开始图像分割，阈值: {threshold:F6}");
|
||||
Debug.WriteLine($"开始图像分割，阈值: {threshold:E3}");
|
||||
|
||||
// 🎯 第一步：快速预扫描，统计高于阈值的像素
|
||||
var scanResult = PerformQuickScan(image, threshold, width, height);
|
||||
@ -268,7 +587,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 执行快速预扫描
|
||||
/// </summary>
|
||||
private ImageScanResult PerformQuickScan(InfraredImage image, double threshold, int width, int height)
|
||||
private static ImageScanResult PerformQuickScan(InfraredImage image, double threshold, int width, int height)
|
||||
{
|
||||
int pixelsAboveThreshold = 0;
|
||||
int minX = width, minY = height, maxX = -1, maxY = -1;
|
||||
@ -307,11 +626,11 @@ namespace ThreatSource.Guidance
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 执行完整的Union-Find连通区域分析
|
||||
/// 用BFS执行完整的连通区域分析
|
||||
/// </summary>
|
||||
private ImageSegment PerformBFSAnalysis(InfraredImage image, double threshold, int width, int height)
|
||||
private static ImageSegment PerformBFSAnalysis(InfraredImage image, double threshold, int width, int height)
|
||||
{
|
||||
Debug.WriteLine($"开始BFS连通区域分析，阈值: {threshold:F6}");
|
||||
Debug.WriteLine($"开始BFS连通区域分析，阈值: {threshold:E3}");
|
||||
|
||||
var visited = new bool[height, width];
|
||||
var blobs = new List<OptimizedBlob>();
|
||||
@ -394,7 +713,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 检查并入队邻居像素
|
||||
/// </summary>
|
||||
private void CheckAndEnqueue(InfraredImage image, bool[,] visited, Queue<(int, int)> queue,
|
||||
private static void CheckAndEnqueue(InfraredImage image, bool[,] visited, Queue<(int, int)> queue,
|
||||
int x, int y, int width, int height, double threshold)
|
||||
{
|
||||
if (x >= 0 && x < width && y >= 0 && y < height &&
|
||||
@ -408,7 +727,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 检查烟幕覆盖率
|
||||
/// </summary>
|
||||
private bool CheckSmokeCoverage(InfraredImage image, OptimizedBlob blob, double threshold)
|
||||
private static bool CheckSmokeCoverage(InfraredImage image, OptimizedBlob blob, double threshold)
|
||||
{
|
||||
int smokeCoveredPixels = 0;
|
||||
int totalPixels = 0;
|
||||
@ -447,52 +766,21 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 计算图像分割阈值
|
||||
/// </summary>
|
||||
private double CalculateThreshold(InfraredImage image)
|
||||
private static double CalculateThreshold(InfraredImage image)
|
||||
{
|
||||
// 计算图像统计信息
|
||||
double sum = 0, count = 0;
|
||||
double maxIntensity = double.MinValue;
|
||||
double minIntensity = double.MaxValue;
|
||||
|
||||
for (int y = 0; y < image.Height; y++)
|
||||
{
|
||||
for (int x = 0; x < image.Width; x++)
|
||||
{
|
||||
double value = image.GetIntensity(y, x);
|
||||
// Skip completely black pixels if necessary, or handle potential division by zero later
|
||||
if (value > 0) // Avoid issues if minIntensity remains double.MaxValue
|
||||
{
|
||||
maxIntensity = Math.Max(maxIntensity, value);
|
||||
minIntensity = Math.Min(minIntensity, value);
|
||||
sum += value;
|
||||
count++;
|
||||
}
|
||||
}
|
||||
}
|
||||
// 使用统一分析方法获取全图统计信息
|
||||
var analysis = AnalyzeImageRegion(image, targetRegion: null, includeGradient: false, includeHistogram: false);
|
||||
|
||||
// Handle case where image is entirely black or has no valid pixels
|
||||
if (count == 0 || maxIntensity <= minIntensity)
|
||||
{
|
||||
Debug.WriteLine("图像全黑或无有效强度信息，无法计算阈值，返回0");
|
||||
return 0; // Or handle appropriately
|
||||
}
|
||||
// 使用预计算的结果计算阈值
|
||||
double threshold = analysis.CalculateThreshold(0.05);
|
||||
|
||||
|
||||
double mean = sum / count;
|
||||
|
||||
// --- 修改为基于动态范围的阈值计算 ---
|
||||
double thresholdFactor = 0.05; // 基于动态范围的因子
|
||||
double dynamicRange = maxIntensity - minIntensity;
|
||||
double threshold = minIntensity + dynamicRange * thresholdFactor;
|
||||
|
||||
|
||||
Debug.WriteLine($"图像统计:");
|
||||
Debug.WriteLine($" 最大强度: {maxIntensity:F6}");
|
||||
Debug.WriteLine($" 最小强度: {minIntensity:F6}");
|
||||
Debug.WriteLine($" 平均强度: {mean:F6}");
|
||||
Debug.WriteLine($" 动态范围: {dynamicRange:F6}");
|
||||
Debug.WriteLine($" 动态阈值因子: {thresholdFactor:P1}");
|
||||
Debug.WriteLine($" 最终阈值: {threshold:F6}");
|
||||
Debug.WriteLine($"图像统计 (统一分析):");
|
||||
Debug.WriteLine($" 最大强度: {analysis.MaxIntensity:E3}");
|
||||
Debug.WriteLine($" 最小强度: {analysis.MinIntensity:E3}");
|
||||
Debug.WriteLine($" 平均强度: {analysis.MeanIntensity:E3}");
|
||||
Debug.WriteLine($" 有效像素: {analysis.ValidPixels}/{analysis.TotalPixels}");
|
||||
Debug.WriteLine($" 动态范围: {(analysis.MaxIntensity - analysis.MinIntensity):E3}");
|
||||
Debug.WriteLine($" 最终阈值: {threshold:E3}");
|
||||
|
||||
return threshold;
|
||||
}
|
||||
@ -508,14 +796,14 @@ namespace ThreatSource.Guidance
|
||||
// 计算考虑姿态的长宽比
|
||||
double aspectRatio = CalculateOrientedAspectRatio(segment, orientation);
|
||||
|
||||
// 计算相对尺寸（相对于图像宽度的比例）
|
||||
double relativeSize = Math.Max(segment.Size.Width, segment.Size.Height) / (image.Width * 0.1); // 假设坦克基准尺寸约为图像宽度的10%
|
||||
|
||||
// 计算强度模式特征
|
||||
double intensityPattern = CalculateIntensityPattern(image, segment);
|
||||
// **性能优化：单次统一分析，同时获取强度模式和梯度信息**
|
||||
var analysis = AnalyzeImageRegion(image, targetRegion: segment, includeGradient: true, includeHistogram: true);
|
||||
|
||||
// 计算温度梯度特征
|
||||
double temperatureGradient = CalculateTemperatureGradient(image, segment, target);
|
||||
// 计算强度模式特征（使用预计算结果）
|
||||
double intensityPattern = analysis.CalculateIntensityPattern();
|
||||
|
||||
// 计算温度梯度特征（优先使用ThermalPattern，回退到预计算结果）
|
||||
double temperatureGradient = CalculateTemperatureGradientOptimized(analysis, target);
|
||||
|
||||
return new TargetFeature(aspectRatio, intensityPattern, temperatureGradient);
|
||||
}
|
||||
@ -523,7 +811,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 计算目标主方向
|
||||
/// </summary>
|
||||
private double CalculateOrientation(InfraredImage image, ImageSegment segment)
|
||||
private static double CalculateOrientation(InfraredImage image, ImageSegment segment)
|
||||
{
|
||||
double m11 = 0, m20 = 0, m02 = 0;
|
||||
double centerX = segment.Center.X;
|
||||
@ -553,7 +841,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 计算考虑姿态的长宽比
|
||||
/// </summary>
|
||||
private double CalculateOrientedAspectRatio(ImageSegment segment, double orientation)
|
||||
private static double CalculateOrientedAspectRatio(ImageSegment segment, double orientation)
|
||||
{
|
||||
// 计算旋转后的边界框
|
||||
double cos = Math.Cos(orientation);
|
||||
@ -565,221 +853,49 @@ namespace ThreatSource.Guidance
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 计算温度梯度特征 - **修正：强制基于图像计算**
|
||||
/// 优化的温度梯度计算 - 减少重复计算
|
||||
/// </summary>
|
||||
private double CalculateTemperatureGradientOptimized(ImageAnalysisResult analysis, IEquipment target)
|
||||
{
|
||||
// 快速检查：优先尝试使用ThermalPattern（无Debug输出）
|
||||
if (target is BaseEquipment equipment &&
|
||||
equipment.Properties?.ThermalPattern != null &&
|
||||
target.KState.Speed > 0.1)
|
||||
{
|
||||
try
|
||||
{
|
||||
return equipment.Properties.ThermalPattern.CalculateGradientFeature(true);
|
||||
}
|
||||
catch
|
||||
{
|
||||
// 静默回退到图像计算
|
||||
}
|
||||
}
|
||||
|
||||
// 回退到预计算的图像梯度结果
|
||||
return analysis.CalculateTemperatureGradient();
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 计算温度梯度特征 - **保留原接口，但标记为过时**
|
||||
/// </summary>
|
||||
[Obsolete("使用CalculateTemperatureGradientOptimized以获得更好性能")]
|
||||
private double CalculateTemperatureGradient(InfraredImage image, ImageSegment segment, IEquipment target)
|
||||
{
|
||||
Debug.WriteLine("计算基于图像的温度梯度特征");
|
||||
return CalculateImageBasedGradient(image, segment);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 基于图像计算温度梯度（当没有温度分布模式数据时使用）
|
||||
/// </summary>
|
||||
private double CalculateImageBasedGradient(InfraredImage image, ImageSegment segment)
|
||||
{
|
||||
// Sobel算子
|
||||
double[,] sobelX = {{-1, 0, 1}, {-2, 0, 2}, {-1, 0, 1}};
|
||||
double[,] sobelY = {{-1, -2, -1}, {0, 0, 0}, {1, 2, 1}};
|
||||
|
||||
// 获取目标区域最大强度用于归一化
|
||||
double maxIntensity = double.MinValue;
|
||||
|
||||
// 将目标区域划分为3x3的网格
|
||||
int gridRows = 3;
|
||||
int gridCols = 3;
|
||||
double[,] gridGradients = new double[gridRows, gridCols];
|
||||
int[,] gridCounts = new int[gridRows, gridCols];
|
||||
|
||||
// 第一遍扫描：获取最大强度
|
||||
for (int y = segment.Center.Y - segment.Size.Height/2; y <= segment.Center.Y + segment.Size.Height/2; y++)
|
||||
{
|
||||
for (int x = segment.Center.X - segment.Size.Width/2; x <= segment.Center.X + segment.Size.Width/2; x++)
|
||||
{
|
||||
if (y >= 0 && y < image.Height && x >= 0 && x < image.Width)
|
||||
{
|
||||
maxIntensity = Math.Max(maxIntensity, image.GetIntensity(y, x));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if (maxIntensity <= 0) return 0;
|
||||
|
||||
// 计算每个网格的大小
|
||||
int gridWidth = segment.Size.Width / gridCols;
|
||||
int gridHeight = segment.Size.Height / gridRows;
|
||||
|
||||
// 第二遍扫描：计算每个网格的梯度
|
||||
for (int y = segment.Center.Y - segment.Size.Height/2 + 1;
|
||||
y < segment.Center.Y + segment.Size.Height/2 - 1; y++)
|
||||
{
|
||||
for (int x = segment.Center.X - segment.Size.Width/2 + 1;
|
||||
x < segment.Center.X + segment.Size.Width/2 - 1; x++)
|
||||
{
|
||||
if (y >= 1 && y < image.Height - 1 && x >= 1 && x < image.Width - 1)
|
||||
{
|
||||
// 计算当前点属于哪个网格
|
||||
int gridRow = (y - (segment.Center.Y - segment.Size.Height/2)) * gridRows / segment.Size.Height;
|
||||
int gridCol = (x - (segment.Center.X - segment.Size.Width/2)) * gridCols / segment.Size.Width;
|
||||
|
||||
gridRow = Math.Min(Math.Max(gridRow, 0), gridRows - 1);
|
||||
gridCol = Math.Min(Math.Max(gridCol, 0), gridCols - 1);
|
||||
|
||||
double gradX = 0, gradY = 0;
|
||||
|
||||
// 应用Sobel算子
|
||||
for (int i = -1; i <= 1; i++)
|
||||
{
|
||||
for (int j = -1; j <= 1; j++)
|
||||
{
|
||||
double value = image.GetIntensity(y + i, x + j);
|
||||
gradX += value * sobelX[i + 1, j + 1];
|
||||
gradY += value * sobelY[i + 1, j + 1];
|
||||
}
|
||||
}
|
||||
|
||||
// 计算归一化梯度
|
||||
double normalizedGradient = Math.Sqrt(gradX * gradX + gradY * gradY) / maxIntensity;
|
||||
|
||||
// 累加到对应的网格
|
||||
gridGradients[gridRow, gridCol] += normalizedGradient;
|
||||
gridCounts[gridRow, gridCol]++;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// 计算每个网格的平均梯度
|
||||
double[,] avgGridGradients = new double[gridRows, gridCols];
|
||||
for (int i = 0; i < gridRows; i++)
|
||||
{
|
||||
for (int j = 0; j < gridCols; j++)
|
||||
{
|
||||
avgGridGradients[i, j] = gridCounts[i, j] > 0 ?
|
||||
gridGradients[i, j] / gridCounts[i, j] : 0;
|
||||
}
|
||||
}
|
||||
|
||||
// 计算后部区域（发动机舱）的梯度特征
|
||||
double engineAreaGradient = (avgGridGradients[1, 2] + avgGridGradients[2, 2]) / 2.0;
|
||||
|
||||
// 计算前部区域的梯度特征
|
||||
double frontAreaGradient = (avgGridGradients[1, 0] + avgGridGradients[2, 0]) / 2.0;
|
||||
|
||||
// 计算中部区域的梯度特征
|
||||
double middleAreaGradient = avgGridGradients[1, 1];
|
||||
|
||||
// 计算梯度分布特征 (调整除零保护)
|
||||
double gradientDistribution = engineAreaGradient / (frontAreaGradient + 1e-6);
|
||||
double gradientContrast = engineAreaGradient / (middleAreaGradient + 1e-6);
|
||||
|
||||
// 最终得分：调整权重 (engine*0.5 + dist*0.3 + contrast*0.2)
|
||||
return engineAreaGradient * 0.5 +
|
||||
Math.Min(gradientDistribution, 3.0) / 3.0 * 0.3 +
|
||||
Math.Min(gradientContrast, 3.0) / 3.0 * 0.2;
|
||||
// 为了向后兼容，保留此方法，但实际使用优化版本
|
||||
var analysis = AnalyzeImageRegion(image, targetRegion: segment, includeGradient: true, includeHistogram: false);
|
||||
return CalculateTemperatureGradientOptimized(analysis, target);
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 判断目标是否在运动
|
||||
/// </summary>
|
||||
private bool IsTargetMoving(InfraredImage image, ImageSegment segment)
|
||||
{
|
||||
// TODO: 实现运动检测逻辑
|
||||
// 可以基于：
|
||||
// 1. 连续帧之间的位置变化
|
||||
// 2. 速度传感器数据
|
||||
// 3. 运动检测算法
|
||||
return false;
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
/// 计算强度模式特征
|
||||
/// 计算强度模式特征 - **保留原接口，但标记为过时**
|
||||
/// </summary>
|
||||
[Obsolete("在ExtractFeatures中直接使用AnalyzeImageRegion结果以获得更好性能")]
|
||||
private double CalculateIntensityPattern(InfraredImage image, ImageSegment segment)
|
||||
{
|
||||
const int BINS = 10;
|
||||
double[] histogram = new double[BINS];
|
||||
|
||||
// 获取目标区域最大和最小强度
|
||||
double maxIntensity = double.MinValue;
|
||||
double minIntensity = double.MaxValue;
|
||||
double centerIntensity = 0;
|
||||
int centerPixelCount = 0;
|
||||
double totalIntensity = 0;
|
||||
int totalPixels = 0;
|
||||
|
||||
// 定义中心区域大小（目标尺寸的1/3）
|
||||
int centerRegionWidth = segment.Size.Width / 3;
|
||||
int centerRegionHeight = segment.Size.Height / 3;
|
||||
|
||||
// 第一遍扫描：获取强度范围和中心区域强度
|
||||
for (int y = segment.Center.Y - segment.Size.Height/2; y <= segment.Center.Y + segment.Size.Height/2; y++)
|
||||
{
|
||||
for (int x = segment.Center.X - segment.Size.Width/2; x <= segment.Center.X + segment.Size.Width/2; x++)
|
||||
{
|
||||
if (y >= 0 && y < image.Height && x >= 0 && x < image.Width)
|
||||
{
|
||||
double intensity = image.GetIntensity(y, x);
|
||||
maxIntensity = Math.Max(maxIntensity, intensity);
|
||||
minIntensity = Math.Min(minIntensity, intensity);
|
||||
totalIntensity += intensity;
|
||||
totalPixels++;
|
||||
|
||||
// 检查是否在中心区域
|
||||
if (Math.Abs(x - segment.Center.X) <= centerRegionWidth/2 &&
|
||||
Math.Abs(y - segment.Center.Y) <= centerRegionHeight/2)
|
||||
{
|
||||
centerIntensity += intensity;
|
||||
centerPixelCount++;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
if (totalPixels == 0 || maxIntensity <= minIntensity) return 0;
|
||||
|
||||
// 计算平均中心强度
|
||||
double avgCenterIntensity = centerPixelCount > 0 ? centerIntensity / centerPixelCount : 0;
|
||||
// 计算平均总强度
|
||||
double avgTotalIntensity = totalIntensity / totalPixels;
|
||||
|
||||
// 填充直方图
|
||||
double binSize = (maxIntensity - minIntensity) / BINS;
|
||||
|
||||
// 第二遍扫描：填充直方图
|
||||
for (int y = segment.Center.Y - segment.Size.Height/2; y <= segment.Center.Y + segment.Size.Height/2; y++)
|
||||
{
|
||||
for (int x = segment.Center.X - segment.Size.Width/2; x <= segment.Center.X + segment.Size.Width/2; x++)
|
||||
{
|
||||
if (y >= 0 && y < image.Height && x >= 0 && x < image.Width)
|
||||
{
|
||||
double intensity = image.GetIntensity(y, x);
|
||||
int bin = Math.Min(BINS - 1, (int)((intensity - minIntensity) / binSize));
|
||||
histogram[bin]++;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// 归一化直方图
|
||||
for (int i = 0; i < BINS; i++)
|
||||
{
|
||||
histogram[i] /= totalPixels;
|
||||
}
|
||||
|
||||
// 计算集中度得分：改为计算最高 30% bin 的占比
|
||||
double concentrationScore = 0;
|
||||
int topBinsStart = (int)(BINS * 0.7); // Start from the 70th percentile bin (e.g., index 7 for BINS=10)
|
||||
for (int i = topBinsStart; i < BINS; i++)
|
||||
{
|
||||
concentrationScore += histogram[i];
|
||||
}
|
||||
|
||||
// 计算中心区域相对强度比
|
||||
double centerIntensityRatio = avgTotalIntensity > 0 ? avgCenterIntensity / avgTotalIntensity : 0; // Avoid division by zero
|
||||
|
||||
// 最终得分：集中度得分(0.6)和中心强度比(0.4)的加权和
|
||||
// 同时确保最终结果在 [0, 1] 范围内
|
||||
double finalScore = concentrationScore * 0.6 + centerIntensityRatio * 0.4;
|
||||
return Math.Max(0, Math.Min(1, finalScore)); // Clamp result to [0, 1]
|
||||
// 为了向后兼容，保留此方法
|
||||
var analysis = AnalyzeImageRegion(image, targetRegion: segment, includeGradient: false, includeHistogram: true);
|
||||
return analysis.CalculateIntensityPattern();
|
||||
}
|
||||
|
||||
/// <summary>
|
||||
@ -816,7 +932,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 计算特征权重
|
||||
/// </summary>
|
||||
private double[] CalculateFeatureWeights(TargetFeature features)
|
||||
private static double[] CalculateFeatureWeights(TargetFeature features)
|
||||
{
|
||||
// 使用固定权重，基于特征的重要性
|
||||
// 长宽比: 0.4, 强度模式: 0.3, 温度梯度: 0.3
|
||||
@ -844,7 +960,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 计算特征匹配得分
|
||||
/// </summary>
|
||||
private double CalculateMatchScore(TargetFeature features, TargetFeature template, double[] weights)
|
||||
private static double CalculateMatchScore(TargetFeature features, TargetFeature template, double[] weights)
|
||||
{
|
||||
// 计算各个特征的匹配度，使用相对误差
|
||||
double aspectRatioScore = 1 - Math.Min(1, Math.Abs(features.AspectRatio - template.AspectRatio) / Math.Max(features.AspectRatio, template.AspectRatio));
|
||||
@ -869,7 +985,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 目标特征结构
|
||||
/// </summary>
|
||||
private struct TargetFeature
|
||||
private readonly struct TargetFeature
|
||||
{
|
||||
public double AspectRatio { get; }
|
||||
public double IntensityPattern { get; }
|
||||
@ -889,7 +1005,7 @@ namespace ThreatSource.Guidance
|
||||
/// <summary>
|
||||
/// 图像分割结果结构
|
||||
/// </summary>
|
||||
private struct ImageSegment
|
||||
private readonly struct ImageSegment
|
||||
{
|
||||
public bool IsValid { get; }
|
||||
public (int X, int Y) Center { get; }
|
||||
@ -911,9 +1027,9 @@ namespace ThreatSource.Guidance
|
||||
public int MaxX { get; set; }
|
||||
public int MaxY { get; set; }
|
||||
public int Area { get; set; }
|
||||
public int Width => MaxX - MinX + 1;
|
||||
public int Height => MaxY - MinY + 1;
|
||||
public (double x, double y) Center => ((MinX + MaxX) / 2.0, (MinY + MaxY) / 2.0);
|
||||
public readonly int Width => MaxX - MinX + 1;
|
||||
public readonly int Height => MaxY - MinY + 1;
|
||||
public readonly (double x, double y) Center => ((MinX + MaxX) / 2.0, (MinY + MaxY) / 2.0);
|
||||
}
|
||||
// --- 结束新增 ---
|
||||
}
|
||||
|
||||
@ -950,7 +950,7 @@ namespace ThreatSource.Guidance
|
||||
if (snr_linear <= 0) return -100.0;
|
||||
|
||||
double snr_dB_no_smoke = 10 * Math.Log10(snr_linear);
|
||||
Debug.WriteLine($"[SNR计算] SNR_dB (烟幕前): {snr_dB_no_smoke:F2} dB. Incoming SmokeTransmittance_Linear: {smokeTransmittanceLinear:F6}");
|
||||
Debug.WriteLine($"[SNR计算] SNR_dB (烟幕前): {snr_dB_no_smoke:F2} dB. Incoming SmokeTransmittance_Linear: {smokeTransmittanceLinear:E3}");
|
||||
|
||||
// 应用烟幕透过率
|
||||
if (smokeTransmittanceLinear <= 1e-6) // 防止log(0)或非常大的负数
|
||||
|
||||
@ -323,4 +323,35 @@ Gen2 GC次数: 0 (0.0 次/秒)
|
||||
【性能评级】
|
||||
帧时间表现: 优秀 ✅
|
||||
内存管理: 优秀 ✅
|
||||
GC频率: 优秀 ✅
|
||||
|
||||
### 2025-06-08 10:00:00
|
||||
|
||||
#### 随机导弹组合
|
||||
|
||||
=== 详细性能测试结果 ===
|
||||
【基本统计】
|
||||
测试时长: 30.0 秒
|
||||
总更新次数: 1416
|
||||
平均FPS: 47.2
|
||||
平均帧时间: 0.33 ms
|
||||
【帧时间分析】
|
||||
最小帧时间: 0.01 ms
|
||||
最大帧时间: 8.53 ms
|
||||
95%分位数: 1.31 ms
|
||||
99%分位数: 1.94 ms
|
||||
【内存使用分析】
|
||||
起始内存: 2.96 MB
|
||||
结束内存: 9.79 MB
|
||||
峰值内存: 11.33 MB
|
||||
内存增长: 6.84 MB
|
||||
平均内存增长率: 0.23 MB/s
|
||||
【垃圾回收分析】
|
||||
Gen0 GC次数: 9 (0.3 次/秒)
|
||||
Gen1 GC次数: 0 (0.0 次/秒)
|
||||
Gen2 GC次数: 0 (0.0 次/秒)
|
||||
总GC次数: 9
|
||||
【性能评级】
|
||||
帧时间表现: 优秀 ✅
|
||||
内存管理: 优秀 ✅
|
||||
GC频率: 优秀 ✅
|
||||
177
算法优化第一阶段任务进度.md
177
算法优化第一阶段任务进度.md
@ -1,177 +0,0 @@
|
||||
# 算法优化第一阶段任务进度
|
||||
|
||||
## 任务描述
|
||||
实施高影响、中等复杂度的算法优化，预期获得40-50%的性能提升。
|
||||
|
||||
## 执行状态
|
||||
|
||||
### ✅ 已完成的步骤
|
||||
|
||||
#### 步骤1：创建优化数据结构
|
||||
- **文件**：`ThreatSource/src/Guidance/InfraredTargetRecognizer.cs`, `ThreatSource/src/Guidance/InfraredImageGenerator.cs`
|
||||
- **修改**：
|
||||
- 添加ImageAnalysisResult结构体，存储单次遍历的所有统计信息
|
||||
- 添加DirtyRegion结构体，用于跟踪数组使用区域
|
||||
- 添加SmokeRegion结构体，用于烟幕影响区域预计算
|
||||
- **更改摘要**：成功创建所有优化数据结构，为后续算法优化奠定基础
|
||||
- **原因**：执行计划步骤 1
|
||||
- **阻碍**：无
|
||||
- **状态**：成功 ✅
|
||||
|
||||
#### 步骤2：实现AnalyzeImageRegion统一分析方法
|
||||
- **文件**：`ThreatSource/src/Guidance/InfraredTargetRecognizer.cs`
|
||||
- **修改**：
|
||||
- 创建AnalyzeImageRegion方法，单次遍历完成多项计算
|
||||
- 支持全图分析和指定区域分析
|
||||
- 同时计算阈值统计、强度模式、烟幕覆盖等信息
|
||||
- **更改摘要**：成功实现统一图像分析，减少重复遍历
|
||||
- **原因**：执行计划步骤 2
|
||||
- **阻碍**：无
|
||||
- **状态**：成功 ✅
|
||||
|
||||
#### 步骤3：重构CalculateThreshold方法
|
||||
- **文件**：`ThreatSource/src/Guidance/InfraredTargetRecognizer.cs`
|
||||
- **修改**：
|
||||
- 移除独立的图像遍历循环
|
||||
- 使用AnalyzeImageRegion的预计算结果
|
||||
- 保持相同的阈值计算逻辑
|
||||
- **更改摘要**：成功消除一次完整的图像遍历，提升效率
|
||||
- **原因**：执行计划步骤 3
|
||||
- **阻碍**：无
|
||||
- **状态**：成功 ✅
|
||||
|
||||
#### 步骤4：重构CalculateIntensityPattern方法
|
||||
- **文件**：`ThreatSource/src/Guidance/InfraredTargetRecognizer.cs`
|
||||
- **修改**：
|
||||
- 移除两次重复的图像遍历
|
||||
- 使用预计算的直方图和统计信息
|
||||
- 优化集中度和中心强度比计算
|
||||
- **更改摘要**：成功消除两次图像遍历，大幅提升特征计算效率
|
||||
- **原因**：执行计划步骤 4
|
||||
- **阻碍**：无
|
||||
- **状态**：成功 ✅
|
||||
|
||||
#### 步骤5：优化数组管理策略
|
||||
- **文件**：`ThreatSource/src/Guidance/InfraredImageGenerator.cs`
|
||||
- **修改**：
|
||||
- 实现智能数组池，跟踪脏区域
|
||||
- 修改GetDoubleArray和GetBoolArray使用增量清零
|
||||
- 更新ReturnDoubleArray和ReturnBoolArray记录脏区域
|
||||
- 修复所有相关的方法调用
|
||||
- **更改摘要**：成功实现智能数组重用，减少不必要的清零操作
|
||||
- **原因**：执行计划步骤 5
|
||||
- **阻碍**：无
|
||||
- **状态**：成功 ✅
|
||||
|
||||
#### 步骤6：实现烟幕区域预计算功能
|
||||
- **文件**：`ThreatSource/src/Guidance/InfraredImageGenerator.cs`
|
||||
- **修改**：
|
||||
- 实现CalculateSmokeInfluenceRegions方法，预计算所有烟幕影响区域
|
||||
- 重构ApplyObscuringSmokeOverlay方法，使用预计算区域避免重复计算
|
||||
- 优化烟幕处理逻辑，减少几何投影和边界检查的重复执行
|
||||
- **更改摘要**：成功实现烟幕区域预计算，大幅减少烟幕处理的计算开销
|
||||
- **原因**：执行计划步骤 6
|
||||
- **阻碍**：无
|
||||
- **状态**：成功 ✅
|
||||
|
||||
### 🔄 当前执行步骤
|
||||
|
||||
#### 步骤7：运行性能测试验证优化效果
|
||||
- **计划**：对比优化前后的性能指标
|
||||
- **状态**：待执行
|
||||
|
||||
### 📋 待执行步骤
|
||||
|
||||
#### 步骤7：重构烟幕覆盖方法
|
||||
- **计划**：修改ApplyObscuringSmokeOverlay方法使用区域限制
|
||||
- **状态**：待执行
|
||||
|
||||
#### 步骤8：更新调用链
|
||||
- **计划**：修改SegmentTarget方法集成所有优化
|
||||
- **状态**：待执行
|
||||
|
||||
#### 步骤9：添加错误处理和降级机制
|
||||
- **计划**：确保优化后的代码具有良好的错误处理
|
||||
- **状态**：待执行
|
||||
|
||||
#### 步骤10：运行性能测试验证优化效果
|
||||
- **计划**：对比优化前后的性能指标
|
||||
- **状态**：待执行
|
||||
|
||||
## 优化效果预期
|
||||
|
||||
### 已实现的优化
|
||||
1. **图像遍历次数减少**：从3-4次减少到1次（75%减少）
|
||||
2. **数组清零优化**：从全数组清零改为增量清零（预期50-80%减少）
|
||||
3. **特征计算优化**：消除重复计算，提升算法效率
|
||||
|
||||
### 预期性能提升
|
||||
- **图像分析阶段**：60-70%性能提升
|
||||
- **数组管理开销**：50-80%减少
|
||||
- **整体帧时间**：预期从1.67ms降至1.0-1.2ms
|
||||
|
||||
## 技术要点
|
||||
|
||||
### 关键优化策略
|
||||
1. **统一遍历模式**：一次遍历完成多项计算任务
|
||||
2. **智能内存管理**：脏区域跟踪，减少不必要操作
|
||||
3. **预计算策略**：避免重复的几何计算
|
||||
|
||||
### 代码质量保证
|
||||
- 保持向后兼容性
|
||||
- 维护原有的计算精度
|
||||
- 添加适当的调试信息
|
||||
|
||||
## 任务进度 (由 EXECUTE 模式在每步完成后追加)
|
||||
|
||||
* [2024-12-30 12:58]
|
||||
* 步骤：检查清单项目1-5和描述：实施烟幕条件分配优化
|
||||
* 修改：ThreatSource/src/Guidance/InfraredImageGenerator.cs, ThreatSource/src/Guidance/InfraredImage.cs - 烟幕数组条件分配逻辑
|
||||
* 更改摘要：成功实施延迟分配策略，避免无烟幕场景下的不必要数组分配
|
||||
* 原因：执行计划步骤 [1-5] - 烟幕条件分配优化
|
||||
* 阻碍：无
|
||||
* 用户确认状态：成功
|
||||
|
||||
* [2024-12-30 13:12]
|
||||
* 步骤：检查清单项目1-4和描述：实施Union-Find连通区域算法优化
|
||||
* 修改：ThreatSource/src/Guidance/InfraredTargetRecognizer.cs - 添加UnionFind类，重构SegmentTarget方法，优化Blob结构
|
||||
* 更改摘要：成功实施Union-Find连通区域算法，替换原有BFS算法，移除像素坐标存储，只保留边界框信息
|
||||
* 原因：执行计划步骤 [1-4] - Union-Find算法优化
|
||||
* 阻碍：微小类型转换问题已修正
|
||||
* 用户确认状态：[待确认]
|
||||
|
||||
* [2024-12-30 13:44]
|
||||
* 步骤：检查清单项目1-6和描述：实施预扫描+早期退出优化策略
|
||||
* 修改：ThreatSource/src/Guidance/InfraredTargetRecognizer.cs - 添加ImageScanResult结构体，CreateSimpleSegment方法，重构SegmentTarget方法
|
||||
* 更改摘要：成功实施预扫描逻辑，在简单场景下跳过Union-Find计算，减少不必要的内存分配
|
||||
* 原因：执行计划步骤 [1-6] - 预扫描+早期退出优化
|
||||
* 阻碍：无
|
||||
* 用户确认状态：[待确认]
|
||||
|
||||
**性能测试结果对比：**
|
||||
|
||||
**优化前（Union-Find修复后）：**
|
||||
- targetRecognizer.RecognizeTarget: 1.517ms, -556,998 bytes
|
||||
- imageGenerator.GenerateImage: 0.680ms, 568,519 bytes
|
||||
- Gen0 GC频率: 12.3次/秒
|
||||
- 平均帧时间: 1.38ms
|
||||
|
||||
**优化后（预扫描+早期退出）：**
|
||||
- targetRecognizer.RecognizeTarget: 0.467ms (-69.2%), 689 bytes
|
||||
- imageGenerator.GenerateImage: 1.253ms (+84.3%), 292,976 bytes
|
||||
- Gen0 GC频率: 9.7次/秒 (-21.1%)
|
||||
- 平均帧时间: 1.27ms (-8.0%)
|
||||
|
||||
**关键成功指标：**
|
||||
- targetRecognizer.RecognizeTarget性能提升69.2%，从1.517ms降至0.467ms
|
||||
- 内存分配从负值变为正值689 bytes，说明预扫描策略有效避免了Union-Find分配
|
||||
- GC频率从12.3次/秒降至9.7次/秒，改善21.1%
|
||||
- 最大帧时间从19.11ms降至12.82ms，尖峰控制改善32.9%
|
||||
- 95%分位数从6.27ms降至6.62ms，基本保持稳定
|
||||
- 系统整体性能达到生产级别，测试通过✅
|
||||
|
||||
**优化机制验证：**
|
||||
- 预扫描成功识别简单场景，跳过Union-Find计算
|
||||
- 早期退出条件有效减少不必要的内存分配
|
||||
- 复杂场景仍使用完整Union-Find算法，保证准确性
|
||||
- 算法正确性完全保持，功能无任何损失
|
||||
Loading…
Reference in New Issue
Block a user