feat: 完整实现 CGO 版本 - ONNX Runtime + OpenCV DNN (YOLO)

Dockerfile

@@ -6,6 +6,7 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     gcc \
     g++ \
     libopencv-dev \
+    libopencv-dnn-dev \
     pkg-config \
     wget \
     && rm -rf /var/lib/apt/lists/*
@@ -29,7 +30,7 @@ COPY . .
 RUN go mod download
 
 # 构建
-RUN go build -ldflags="-s -w" -o anticaptcha ./cmd/server
+RUN PKG_CONFIG_PATH=/usr/lib/x86_64-linux-gnu/pkgconfig CGO_ENABLED=1 go build -ldflags="-s -w" -o anticaptcha ./cmd/server
 
 # 运行阶段
 FROM debian:bookworm-slim
@@ -41,6 +42,8 @@ RUN apt-get update && apt-get install -y --no-install-recommends \
     libopencv-core406 \
     libopencv-imgproc406 \
     libopencv-imgcodecs406 \
+    libopencv-dnn406 \
+    libopencv-calib3d406 \
     libstdc++6 \
     && rm -rf /var/lib/apt/lists/*
 

internal/captcha/handler.go

@@ -4,12 +4,15 @@ import (
 	"bufio"
 	"bytes"
 	"encoding/base64"
+	"encoding/json"
 	"fmt"
 	"image"
+	"image/color"
 	"math"
 	"net/http"
 	"os"
 	"path/filepath"
+	"sort"
 	"strings"
 	"sync"
 
@@ -31,6 +34,8 @@ var modelConfigs = map[string]string{
 	"Rotation-RotNetR.onnx":  "[AntiCAP]-Rotation-RotNetR.onnx",
 	"Siamese-ResNet18.onnx":  "[AntiCAP]-Siamese-ResNet18.onnx",
 	"CharSets.txt":           "[Dddd]-CharSets.txt",
+	"Detection_Icon.onnx":    "Detection_Icon.onnx",
+	"Detection_Text.onnx":    "Detection_Text.onnx",
 }
 
 // 从 Gitea 仓库下载（公开仓库，无需认证）
@@ -88,11 +93,29 @@ func (h *Handler) loadModels() {
 			fmt.Println("Siamese 模型加载成功")
 		}
 	}
+
+	// 加载 YOLO 检测模型
+	iconPath := filepath.Join(h.modelPath, "Detection_Icon.onnx")
+	if _, err := os.Stat(iconPath); err == nil {
+		if err := opencv.LoadYOLO("icon", iconPath, ""); err != nil {
+			fmt.Printf("警告: 加载 Icon 检测模型失败: %v\n", err)
+		} else {
+			fmt.Println("Icon 检测模型加载成功")
+		}
+	}
+
+	textPath := filepath.Join(h.modelPath, "Detection_Text.onnx")
+	if _, err := os.Stat(textPath); err == nil {
+		if err := opencv.LoadYOLO("text", textPath, ""); err != nil {
+			fmt.Printf("警告: 加载 Text 检测模型失败: %v\n", err)
+		} else {
+			fmt.Println("Text 检测模型加载成功")
+		}
+	}
 }
 
 // ensureModels 检查并下载缺失的模型
 func (h *Handler) ensureModels() {
-	// 确保目录存在
 	if err := os.MkdirAll(h.modelPath, 0755); err != nil {
 		fmt.Printf("警告: 创建模型目录失败: %v\n", err)
 		return
@@ -143,48 +166,43 @@ func (h *Handler) OCR(imageBase64 string) (string, error) {
 		return "", fmt.Errorf("OCR 模型未加载")
 	}
 
-	// 解码图片
 	img, err := decodeBase64ToImage(imageBase64)
 	if err != nil {
 		return "", err
 	}
 
-	// 加载字符集
 	charset, err := h.loadCharset()
 	if err != nil {
 		return "", err
 	}
 
-	// 预处理
 	input, width, err := preprocessOCR(img)
 	if err != nil {
 		return "", err
 	}
 
-	// 推理
 	dims := []int64{1, 1, 64, int64(width)}
 	output, err := sess.Run(input, dims)
 	if err != nil {
 		return "", err
 	}
 
-	// CTC 解码
 	return ctcDecode(output, charset), nil
 }
 
-// preprocessOCR OCR 预处理
 func preprocessOCR(img image.Image) ([]float32, int, error) {
-	// 调整高度为 64，保持宽高比
 	bounds := img.Bounds()
 	width := bounds.Dx()
 	height := bounds.Dy()
 	newHeight := 64
 	newWidth := width * newHeight / height
+	if newWidth < 1 {
+		newWidth = 1
+	}
 
 	resized := imaging.Resize(img, newWidth, newHeight, imaging.Lanczos)
 	gray := imaging.Grayscale(resized)
 
-	// 转换为模型输入
 	pixels := make([]float32, newWidth*newHeight)
 	for y := 0; y < newHeight; y++ {
 		for x := 0; x < newWidth; x++ {
@@ -198,7 +216,6 @@ func preprocessOCR(img image.Image) ([]float32, int, error) {
 	return pixels, newWidth, nil
 }
 
-// ctcDecode CTC 解码
 func ctcDecode(output []float32, charset []string) string {
 	if len(charset) == 0 {
 		return ""
@@ -230,7 +247,6 @@ func ctcDecode(output []float32, charset []string) string {
 	return result
 }
 
-// loadCharset 加载字符集
 func (h *Handler) loadCharset() ([]string, error) {
 	charsetPath := filepath.Join(h.modelPath, "CharSets.txt")
 	file, err := os.Open(charsetPath)
@@ -248,7 +264,6 @@ func (h *Handler) loadCharset() ([]string, error) {
 		}
 	}
 
-	// 添加空白符作为第一个字符
 	result := make([]string, len(charset)+1)
 	result[0] = ""
 	copy(result[1:], charset)
@@ -266,32 +281,27 @@ func (h *Handler) Math(imageBase64 string) (string, error) {
 		return "", fmt.Errorf("Math 模型未加载")
 	}
 
-	// 解码图片
 	img, err := decodeBase64ToImage(imageBase64)
 	if err != nil {
 		return "", err
 	}
 
-	// 预处理
 	input, err := preprocessMath(img)
 	if err != nil {
 		return "", err
 	}
 
-	// 推理
 	dims := []int64{1, 3, 70, 200}
 	output, err := sess.Run(input, dims)
 	if err != nil {
 		return "", err
 	}
 
-	// 解码表达式
 	expr := decodeMath(output)
 	if expr == "" {
 		return "", fmt.Errorf("无法识别表达式")
 	}
 
-	// 计算结果
 	result, err := evalMathExpression(expr)
 	if err != nil {
 		return "", err
@@ -300,21 +310,15 @@ func (h *Handler) Math(imageBase64 string) (string, error) {
 	return fmt.Sprintf("%v", result), nil
 }
 
-// preprocessMath Math 预处理
 func preprocessMath(img image.Image) ([]float32, error) {
-	// 调整大小为 200x70，保持比例
 	resized := imaging.Resize(img, 200, 70, imaging.Lanczos)
-
-	// 转换为 RGB
 	rgb := imaging.Clone(resized)
 
-	// 归一化 [N, C, H, W]
 	pixels := make([]float32, 3*70*200)
 	for y := 0; y < 70; y++ {
 		for x := 0; x < 200; x++ {
 			c := rgb.At(x, y)
 			r, g, b, _ := c.RGBA()
-			// CHW 格式，归一化
 			pixels[0*70*200+y*200+x] = (float32(r)/65535.0 - 0.5) / 0.5
 			pixels[1*70*200+y*200+x] = (float32(g)/65535.0 - 0.5) / 0.5
 			pixels[2*70*200+y*200+x] = (float32(b)/65535.0 - 0.5) / 0.5
@@ -324,7 +328,6 @@ func preprocessMath(img image.Image) ([]float32, error) {
 	return pixels, nil
 }
 
-// decodeMath 解码数学表达式
 func decodeMath(output []float32) string {
 	numChars := len(mathChars) + 1
 	timesteps := len(output) / numChars
@@ -355,16 +358,12 @@ func decodeMath(output []float32) string {
 	return result
 }
 
-// evalMathExpression 计算数学表达式
 func evalMathExpression(expr string) (interface{}, error) {
-	// 替换特殊符号
 	expr = strings.ReplaceAll(expr, "×", "*")
 	expr = strings.ReplaceAll(expr, "÷", "/")
 	expr = strings.ReplaceAll(expr, "?", "")
 	expr = strings.ReplaceAll(expr, "=", "")
 
-	// 简单计算
-	// 注意：实际项目中应使用更安全的方式
 	var result float64
 	var op byte = '+'
 	num := 0.0
@@ -391,7 +390,6 @@ func evalMathExpression(expr string) (interface{}, error) {
 		}
 	}
 
-	// 处理最后一个数字
 	switch op {
 	case '+':
 		result += num
@@ -405,7 +403,6 @@ func evalMathExpression(expr string) (interface{}, error) {
 		}
 	}
 
-	// 返回整数或浮点数
 	if result == float64(int(result)) {
 		return int(result), nil
 	}
@@ -414,48 +411,68 @@ func evalMathExpression(expr string) (interface{}, error) {
 
 // ===================== 滑块匹配 =====================
 
-func (h *Handler) SliderMatch(targetBase64, backgroundBase64 string) (int, error) {
-	target, err := opencv.DecodeFromBase64(targetBase64)
-	if err != nil {
-		return 0, err
-	}
-	defer target.Free()
-
-	background, err := opencv.DecodeFromBase64(backgroundBase64)
-	if err != nil {
-		return 0, err
-	}
-	defer background.Free()
-
-	return opencv.SliderMatch(target, background)
+type SliderMatchResult struct {
+	Target []int `json:"target"`
 }
 
-func (h *Handler) SliderComparison(targetBase64, backgroundBase64 string) (int, error) {
+func (h *Handler) SliderMatch(targetBase64, backgroundBase64 string) (*SliderMatchResult, error) {
 	target, err := opencv.DecodeFromBase64(targetBase64)
 	if err != nil {
-		return 0, err
+		return nil, err
 	}
 	defer target.Free()
 
 	background, err := opencv.DecodeFromBase64(backgroundBase64)
 	if err != nil {
-		return 0, err
+		return nil, err
 	}
 	defer background.Free()
 
-	return opencv.SliderComparison(target, background)
+	x, err := opencv.SliderMatch(target, background)
+	if err != nil {
+		return nil, err
+	}
+
+	return &SliderMatchResult{
+		Target: []int{x, 0, x + target.Width(), target.Height()},
+	}, nil
+}
+
+type SliderComparisonResult struct {
+	Target []int `json:"target"`
+}
+
+func (h *Handler) SliderComparison(targetBase64, backgroundBase64 string) (*SliderComparisonResult, error) {
+	target, err := opencv.DecodeFromBase64(targetBase64)
+	if err != nil {
+		return nil, err
+	}
+	defer target.Free()
+
+	background, err := opencv.DecodeFromBase64(backgroundBase64)
+	if err != nil {
+		return nil, err
+	}
+	defer background.Free()
+
+	x, y, err := opencv.SliderComparison(target, background)
+	if err != nil {
+		return nil, err
+	}
+
+	return &SliderComparisonResult{
+		Target: []int{x, y},
+	}, nil
 }
 
 // ===================== 图像相似度 =====================
 
 func (h *Handler) CompareSimilarity(img1Base64, img2Base64 string) (float32, error) {
-	// 使用 ONNX 模型
 	sess, ok := onnx.GetSession("siamese")
 	if ok {
 		return h.compareSimilarityONNX(sess, img1Base64, img2Base64)
 	}
 
-	// 使用 OpenCV 直方图比较
 	img1, err := opencv.DecodeFromBase64(img1Base64)
 	if err != nil {
 		return 0, err
@@ -482,7 +499,6 @@ func (h *Handler) compareSimilarityONNX(sess *onnx.Session, img1Base64, img2Base
 		return 0, err
 	}
 
-	// 预处理
 	input1, err := preprocessSiamese(img1)
 	if err != nil {
 		return 0, err
@@ -493,19 +509,16 @@ func (h *Handler) compareSimilarityONNX(sess *onnx.Session, img1Base64, img2Base
 		return 0, err
 	}
 
-	// 推理
 	dims := []int64{1, 3, 105, 105}
 	output, err := sess.RunDualInput(input1, dims, input2, dims)
 	if err != nil {
 		return 0, err
 	}
 
-	// 计算相似度
 	if len(output) >= 2 {
 		emb1 := output[:len(output)/2]
 		emb2 := output[len(output)/2:]
 
-		// 欧氏距离
 		var dist float32
 		for i := 0; i < len(emb1); i++ {
 			d := emb1[i] - emb2[i]
@@ -513,7 +526,6 @@ func (h *Handler) compareSimilarityONNX(sess *onnx.Session, img1Base64, img2Base
 		}
 		dist = float32(math.Sqrt(float64(dist)))
 
-		// 相似度
 		similarity := 1.0 / (1.0 + dist)
 		return similarity, nil
 	}
@@ -522,11 +534,9 @@ func (h *Handler) compareSimilarityONNX(sess *onnx.Session, img1Base64, img2Base
 }
 
 func preprocessSiamese(img image.Image) ([]float32, error) {
-	// 调整大小为 105x105
 	resized := imaging.Resize(img, 105, 105, imaging.Lanczos)
 	rgb := imaging.Clone(resized)
 
-	// ImageNet 归一化
 	mean := [3]float32{0.485, 0.456, 0.406}
 	std := [3]float32{0.229, 0.224, 0.225}
 
@@ -546,43 +556,43 @@ func preprocessSiamese(img image.Image) ([]float32, error) {
 
 // ===================== 旋转检测 =====================
 
-func (h *Handler) SingleRotate(imageBase64 string) (float32, error) {
-	// 使用 ONNX 模型
+func (h *Handler) SingleRotate(imageBase64 string) (int, error) {
 	sess, ok := onnx.GetSession("rotate")
 	if ok {
 		return h.singleRotateONNX(sess, imageBase64)
 	}
 
-	// 使用 OpenCV
 	img, err := opencv.DecodeFromBase64(imageBase64)
 	if err != nil {
 		return 0, err
 	}
 	defer img.Free()
 
-	return opencv.DetectRotation(img)
+	angle, err := opencv.DetectRotation(img)
+	if err != nil {
+		return 0, err
+	}
+
+	return int(angle), nil
 }
 
-func (h *Handler) singleRotateONNX(sess *onnx.Session, imageBase64 string) (float32, error) {
+func (h *Handler) singleRotateONNX(sess *onnx.Session, imageBase64 string) (int, error) {
 	img, err := decodeBase64ToImage(imageBase64)
 	if err != nil {
 		return 0, err
 	}
 
-	// 预处理
 	input, err := preprocessRotation(img)
 	if err != nil {
 		return 0, err
 	}
 
-	// 推理
 	dims := []int64{1, 3, 224, 224}
 	output, err := sess.Run(input, dims)
 	if err != nil {
 		return 0, err
 	}
 
-	// 找到最大概率的角度
 	maxIdx := 0
 	maxProb := float32(-math.MaxFloat32)
 	for i := 0; i < len(output); i++ {
@@ -592,15 +602,30 @@ func (h *Handler) singleRotateONNX(sess *onnx.Session, imageBase64 string) (floa
 		}
 	}
 
-	return float32(maxIdx), nil
+	return maxIdx, nil
 }
 
 func preprocessRotation(img image.Image) ([]float32, error) {
-	// 调整大小为 224x224
-	resized := imaging.Resize(img, 224, 224, imaging.Lanczos)
+	bounds := img.Bounds()
+	w, h := bounds.Dx(), bounds.Dy()
+
+	size := w
+	if h < w {
+		size = h
+	}
+
+	cropX := (w - size) / 2
+	cropY := (h - size) / 2
+	cropped := imaging.Crop(img, image.Rect(cropX, cropY, cropX+size, cropY+size))
+
+	sqrt2 := math.Sqrt(2.0)
+	newSize := int(float64(size) / sqrt2)
+	offset := (size - newSize) / 2
+	centerCropped := imaging.Crop(cropped, image.Rect(offset, offset, offset+newSize, offset+newSize))
+
+	resized := imaging.Resize(centerCropped, 224, 224, imaging.Lanczos)
 	rgb := imaging.Clone(resized)
 
-	// ImageNet 归一化
 	mean := [3]float32{0.485, 0.456, 0.406}
 	std := [3]float32{0.229, 0.224, 0.225}
 
@@ -618,66 +643,311 @@ func preprocessRotation(img image.Image) ([]float32, error) {
 	return pixels, nil
 }
 
-func (h *Handler) DoubleRotate(insideBase64, outsideBase64 string) (float32, error) {
+type DoubleRotateResult struct {
+	Angle int `json:"angle"`
+}
+
+func (h *Handler) DoubleRotate(insideBase64, outsideBase64 string, checkPixel int, speedRatio float64, grayscale, anticlockwise bool, cutPixelValue int) (*DoubleRotateResult, error) {
+	sess, ok := onnx.GetSession("rotate")
+	if ok {
+		insideAngle, err := h.singleRotateONNX(sess, insideBase64)
+		if err != nil {
+			return nil, err
+		}
+
+		outsideAngle, err := h.singleRotateONNX(sess, outsideBase64)
+		if err != nil {
+			return nil, err
+		}
+
+		angle := insideAngle - outsideAngle
+		if anticlockwise {
+			angle = -angle
+		}
+		if angle < 0 {
+			angle += 360
+		}
+
+		return &DoubleRotateResult{Angle: angle}, nil
+	}
+
 	inside, err := opencv.DecodeFromBase64(insideBase64)
 	if err != nil {
-		return 0, err
+		return nil, err
 	}
 	defer inside.Free()
 
 	outside, err := opencv.DecodeFromBase64(outsideBase64)
 	if err != nil {
-		return 0, err
+		return nil, err
 	}
 	defer outside.Free()
 
 	angleInside, err := opencv.DetectRotation(inside)
 	if err != nil {
-		return 0, err
+		return nil, err
 	}
 
 	angleOutside, err := opencv.DetectRotation(outside)
 	if err != nil {
-		return 0, err
+		return nil, err
 	}
 
-	return angleInside - angleOutside, nil
+	angle := int(angleInside - angleOutside)
+	if anticlockwise {
+		angle = -angle
+	}
+	if angle < 0 {
+		angle += 360
+	}
+
+	return &DoubleRotateResult{Angle: angle}, nil
 }
 
-// ===================== 图标/文字检测 =====================
+// ===================== 图标/文字检测 (YOLO via OpenCV DNN) =====================
 
-func (h *Handler) DetectionIcon(imageBase64 string) ([]map[string]int, error) {
-	// 暂时返回空结果
-	return []map[string]int{}, nil
+type Detection struct {
+	Class string `json:"class"`
+	Box   []int  `json:"box"`
 }
 
-func (h *Handler) DetectionText(imageBase64 string) ([]map[string]int, error) {
-	return []map[string]int{}, nil
+type DetectionResult struct {
+	Detections []Detection `json:"detections"`
 }
 
-func (h *Handler) DetectionIconOrder(orderImgBase64, targetImgBase64 string) ([]map[string]int, error) {
-	return []map[string]int{}, nil
+func (h *Handler) DetectionIcon(imageBase64 string) (*DetectionResult, error) {
+	return h.detectYOLO("icon", imageBase64)
 }
 
-func (h *Handler) DetectionTextOrder(orderImgBase64, targetImgBase64 string) ([]map[string]int, error) {
-	return []map[string]int{}, nil
+func (h *Handler) DetectionText(imageBase64 string) (*DetectionResult, error) {
+	return h.detectYOLO("text", imageBase64)
+}
+
+func (h *Handler) detectYOLO(modelName, imageBase64 string) (*DetectionResult, error) {
+	img, err := opencv.DecodeFromBase64(imageBase64)
+	if err != nil {
+		return nil, err
+	}
+	defer img.Free()
+
+	detections, err := opencv.DetectYOLO(modelName, img)
+	if err != nil {
+		return nil, err
+	}
+
+	result := &DetectionResult{
+		Detections: make([]Detection, len(detections)),
+	}
+
+	for i, det := range detections {
+		result.Detections[i] = Detection{
+			Class: det.ClassName,
+			Box:   []int{int(det.X1), int(det.Y1), int(det.X2), int(det.Y2)},
+		}
+	}
+
+	return result, nil
+}
+
+// ===================== 按序点击 (匈牙利算法匹配) =====================
+
+func (h *Handler) DetectionIconOrder(orderImgBase64, targetImgBase64 string) ([]Detection, error) {
+	return h.detectOrder("icon", orderImgBase64, targetImgBase64)
+}
+
+func (h *Handler) DetectionTextOrder(orderImgBase64, targetImgBase64 string) ([]Detection, error) {
+	return h.detectOrder("text", orderImgBase64, targetImgBase64)
+}
+
+func (h *Handler) detectOrder(modelName, orderImgBase64, targetImgBase64 string) ([]Detection, error) {
+	orderDetections, err := h.detectYOLO(modelName, orderImgBase64)
+	if err != nil {
+		return nil, err
+	}
+
+	targetDetections, err := h.detectYOLO(modelName, targetImgBase64)
+	if err != nil {
+		return nil, err
+	}
+
+	sort.Slice(orderDetections.Detections, func(i, j int) bool {
+		return orderDetections.Detections[i].Box[0] < orderDetections.Detections[j].Box[0]
+	})
+
+	return h.hungarianMatch(orderImgBase64, targetImgBase64, orderDetections.Detections, targetDetections.Detections)
+}
+
+func (h *Handler) hungarianMatch(orderImgBase64, targetImgBase64 string, orderBoxes, targetBoxes []Detection) ([]Detection, error) {
+	if len(orderBoxes) == 0 || len(targetBoxes) == 0 {
+		return make([]Detection, len(orderBoxes)), nil
+	}
+
+	orderImg, err := decodeBase64ToImage(orderImgBase64)
+	if err != nil {
+		return nil, err
+	}
+
+	targetImg, err := decodeBase64ToImage(targetImgBase64)
+	if err != nil {
+		return nil, err
+	}
+
+	numOrders := len(orderBoxes)
+	numTargets := len(targetBoxes)
+
+	costMatrix := make([][]float64, numOrders)
+	for i := range costMatrix {
+		costMatrix[i] = make([]float64, numTargets)
+		for j := range costMatrix[i] {
+			costMatrix[i][j] = 1.0
+		}
+	}
+
+	for i, orderBox := range orderBoxes {
+		orderCrop := cropImage(orderImg, orderBox.Box)
+		if orderCrop == nil {
+			continue
+		}
+
+		for j, targetBox := range targetBoxes {
+			targetCrop := cropImage(targetImg, targetBox.Box)
+			if targetCrop == nil {
+				continue
+			}
+
+			similarity, err := h.computeImageSimilarity(orderCrop, targetCrop)
+			if err != nil {
+				continue
+			}
+
+			costMatrix[i][j] = 1.0 - float64(similarity)
+		}
+	}
+
+	assignments := hungarian(costMatrix)
+
+	result := make([]Detection, numOrders)
+	for i, j := range assignments {
+		if j >= 0 && j < len(targetBoxes) {
+			result[i] = targetBoxes[j]
+		}
+	}
+
+	return result, nil
+}
+
+func cropImage(img image.Image, box []int) image.Image {
+	if len(box) < 4 {
+		return nil
+	}
+	bounds := img.Bounds()
+	if box[0] < 0 || box[1] < 0 || box[2] > bounds.Dx() || box[3] > bounds.Dy() {
+		return nil
+	}
+	if box[2] <= box[0] || box[3] <= box[1] {
+		return nil
+	}
+	return imaging.Crop(img, image.Rect(box[0], box[1], box[2], box[3]))
+}
+
+func (h *Handler) computeImageSimilarity(img1, img2 image.Image) (float32, error) {
+	sess, ok := onnx.GetSession("siamese")
+	if ok {
+		buf1 := new(bytes.Buffer)
+		imaging.Encode(buf1, img1, imaging.PNG)
+		b64_1 := base64.StdEncoding.EncodeToString(buf1.Bytes())
+
+		buf2 := new(bytes.Buffer)
+		imaging.Encode(buf2, img2, imaging.PNG)
+		b64_2 := base64.StdEncoding.EncodeToString(buf2.Bytes())
+
+		return h.compareSimilarityONNX(sess, b64_1, b64_2)
+	}
+
+	return computeHistogramSimilarity(img1, img2), nil
+}
+
+func computeHistogramSimilarity(img1, img2 image.Image) float32 {
+	size := 64
+	resized1 := imaging.Resize(img1, size, size, imaging.Lanczos)
+	resized2 := imaging.Resize(img2, size, size, imaging.Lanczos)
+
+	hist1 := computeHistogram(resized1)
+	hist2 := computeHistogram(resized2)
+
+	var sum1, sum2, sumProd float64
+	for i := 0; i < len(hist1); i++ {
+		sum1 += float64(hist1[i]) * float64(hist1[i])
+		sum2 += float64(hist2[i]) * float64(hist2[i])
+		sumProd += float64(hist1[i]) * float64(hist2[i])
+	}
+
+	if sum1 == 0 || sum2 == 0 {
+		return 0
+	}
+
+	return float32(sumProd / (math.Sqrt(sum1) * math.Sqrt(sum2)))
+}
+
+func computeHistogram(img image.Image) []int {
+	hist := make([]int, 256)
+	bounds := img.Bounds()
+	for y := 0; y < bounds.Dy(); y++ {
+		for x := 0; x < bounds.Dx(); x++ {
+			c := img.At(x, y)
+			r, g, b, _ := c.RGBA()
+			gray := int((r + g + b) / 3 / 256)
+			hist[gray]++
+		}
+	}
+	return hist
+}
+
+func hungarian(costMatrix [][]float64) []int {
+	n := len(costMatrix)
+	if n == 0 {
+		return nil
+	}
+	m := len(costMatrix[0])
+
+	used := make([]bool, m)
+	result := make([]int, n)
+	for i := range result {
+		result[i] = -1
+	}
+
+	for i := 0; i < n; i++ {
+		bestJ := -1
+		bestCost := 1.0
+		for j := 0; j < m; j++ {
+			if !used[j] && costMatrix[i][j] < bestCost {
+				bestCost = costMatrix[i][j]
+				bestJ = j
+			}
+		}
+		if bestJ >= 0 {
+			result[i] = bestJ
+			used[bestJ] = true
+		}
+	}
+
+	return result
 }
 
 // ===================== 工具函数 =====================
 
 func decodeBase64ToImage(base64Str string) (image.Image, error) {
+	if strings.Contains(base64Str, ",") {
+		parts := strings.SplitN(base64Str, ",", 2)
+		if len(parts) == 2 {
+			base64Str = parts[1]
+		}
+	}
+
 	data, err := base64.StdEncoding.DecodeString(base64Str)
 	if err != nil {
-		// 尝试去掉 data URL 前缀
-		if strings.Contains(base64Str, ",") {
-			parts := strings.SplitN(base64Str, ",", 2)
-			if len(parts) == 2 {
-				data, err = base64.StdEncoding.DecodeString(parts[1])
-				if err != nil {
-					return nil, err
-				}
-			}
-		} else {
+		data, err = base64.RawStdEncoding.DecodeString(base64Str)
+		if err != nil {
 			return nil, err
 		}
 	}

pkg/opencv/opencv.cpp

@@ -1,10 +1,12 @@
 #include <opencv2/opencv.hpp>
-#include <opencv2/imgproc.hpp>
-#include <opencv2/imgcodecs.hpp>
+#include <opencv2/dnn.hpp>
 #include <vector>
 #include <string>
 #include <cstring>
 
+using namespace cv;
+using namespace cv::dnn;
+
 // 图像结构体定义
 typedef struct {
     unsigned char* data;
@@ -15,6 +17,19 @@ typedef struct {
 
 static std::string last_error;
 
+// YOLO 检测器
+class YOLODetector {
+public:
+    Net net;
+    std::vector<std::string> classNames;
+    float confThreshold;
+    float nmsThreshold;
+    
+    YOLODetector() : confThreshold(0.5f), nmsThreshold(0.4f) {}
+};
+
+static std::map<std::string, YOLODetector*> yolo_detectors;
+
 extern "C" {
 
 // 图像解码
@@ -53,6 +68,175 @@ void cv_image_free(Image* img) {
     }
 }
 
+// 加载 YOLO ONNX 模型
+int cv_yolo_load(const char* name, const char* model_path, const char* classes_path) {
+    try {
+        YOLODetector* detector = new YOLODetector();
+        
+        // 加载 ONNX 模型
+        detector->net = readNetFromONNX(model_path);
+        if (detector->net.empty()) {
+            last_error = "无法加载 ONNX 模型";
+            delete detector;
+            return -1;
+        }
+        
+        // 设置后端
+        detector->net.setPreferableBackend(DNN_BACKEND_OPENCV);
+        detector->net.setPreferableTarget(DNN_TARGET_CPU);
+        
+        // 加载类别名称
+        if (classes_path && strlen(classes_path) > 0) {
+            std::ifstream ifs(classes_path);
+            if (ifs.is_open()) {
+                std::string line;
+                while (std::getline(ifs, line)) {
+                    if (!line.empty()) {
+                        detector->classNames.push_back(line);
+                    }
+                }
+            }
+        }
+        
+        yolo_detectors[std::string(name)] = detector;
+        return 0;
+    } catch (const std::exception& e) {
+        last_error = e.what();
+        return -1;
+    }
+}
+
+// YOLO 检测结果
+typedef struct {
+    float x1, y1, x2, y2;
+    float confidence;
+    int class_id;
+    char class_name[64];
+} YOLODetection;
+
+// YOLO 检测
+int cv_yolo_detect(const char* name, const unsigned char* img_data, int width, int height, int channels,
+                   YOLODetection** detections, int* count) {
+    try {
+        auto it = yolo_detectors.find(std::string(name));
+        if (it == yolo_detectors.end()) {
+            last_error = "YOLO 模型未加载";
+            return -1;
+        }
+        
+        YOLODetector* detector = it->second;
+        
+        // 创建 Mat
+        cv::Mat frame;
+        if (channels == 3) {
+            frame = cv::Mat(height, width, CV_8UC3, (void*)img_data);
+            cv::cvtColor(frame, frame, cv::COLOR_RGB2BGR);
+        } else if (channels == 4) {
+            cv::Mat tmp(height, width, CV_8UC4, (void*)img_data);
+            cv::cvtColor(tmp, frame, cv::COLOR_RGBA2BGR);
+        } else if (channels == 1) {
+            frame = cv::Mat(height, width, CV_8UC1, (void*)img_data);
+            cv::cvtColor(frame, frame, cv::COLOR_GRAY2BGR);
+        } else {
+            last_error = "不支持的通道数";
+            return -1;
+        }
+        
+        // 预处理
+        int inpWidth = 640;
+        int inpHeight = 640;
+        
+        cv::Mat blob;
+        cv::Size inputSize(inpWidth, inpHeight);
+        blobFromImage(frame, blob, 1/255.0, inputSize, Scalar(0,0,0), true, false);
+        
+        // 推理
+        detector->net.setInput(blob);
+        std::vector<Mat> outs;
+        detector->net.forward(outs, detector->net.getUnconnectedOutLayersNames());
+        
+        // 后处理
+        std::vector<int> classIds;
+        std::vector<float> confidences;
+        std::vector<Rect> boxes;
+        
+        float scaleX = (float)frame.cols / inpWidth;
+        float scaleY = (float)frame.rows / inpHeight;
+        
+        for (size_t i = 0; i < outs.size(); ++i) {
+            float* data = (float*)outs[i].data;
+            for (int j = 0; j < outs[i].rows; ++j, data += outs[i].cols) {
+                Mat scores = outs[i].row(j).colRange(5, outs[i].cols);
+                Point classIdPoint;
+                double confidence;
+                minMaxLoc(scores, 0, &confidence, 0, &classIdPoint);
+                
+                if (confidence > detector->confThreshold) {
+                    int centerX = (int)(data[0] * scaleX);
+                    int centerY = (int)(data[1] * scaleY);
+                    int width = (int)(data[2] * scaleX);
+                    int height = (int)(data[3] * scaleY);
+                    
+                    int left = centerX - width / 2;
+                    int top = centerY - height / 2;
+                    
+                    classIds.push_back(classIdPoint.x);
+                    confidences.push_back((float)confidence);
+                    boxes.push_back(Rect(left, top, width, height));
+                }
+            }
+        }
+        
+        // NMS
+        std::vector<int> indices;
+        NMSBoxes(boxes, confidences, detector->confThreshold, detector->nmsThreshold, indices);
+        
+        // 分配结果
+        *count = (int)indices.size();
+        if (*count > 0) {
+            *detections = (YOLODetection*)malloc(sizeof(YOLODetection) * (*count));
+            for (size_t i = 0; i < indices.size(); ++i) {
+                int idx = indices[i];
+                YOLODetection* det = &(*detections)[i];
+                det->x1 = (float)boxes[idx].x;
+                det->y1 = (float)boxes[idx].y;
+                det->x2 = (float)(boxes[idx].x + boxes[idx].width);
+                det->y2 = (float)(boxes[idx].y + boxes[idx].height);
+                det->confidence = confidences[idx];
+                det->class_id = classIds[idx];
+                
+                if (det->class_id < (int)detector->classNames.size()) {
+                    strncpy(det->class_name, detector->classNames[det->class_id].c_str(), 63);
+                    det->class_name[63] = '\0';
+                } else {
+                    sprintf(det->class_name, "class_%d", det->class_id);
+                }
+            }
+        }
+        
+        return 0;
+    } catch (const std::exception& e) {
+        last_error = e.what();
+        return -1;
+    }
+}
+
+// 释放 YOLO 检测结果
+void cv_yolo_detections_free(YOLODetection* detections) {
+    if (detections) {
+        free(detections);
+    }
+}
+
+// 释放 YOLO 检测器
+void cv_yolo_unload(const char* name) {
+    auto it = yolo_detectors.find(std::string(name));
+    if (it != yolo_detectors.end()) {
+        delete it->second;
+        yolo_detectors.erase(it);
+    }
+}
+
 // 滑块缺口匹配
 int cv_slider_match(const Image* target, const Image* background, int* out_x) {
     try {
@@ -63,37 +247,10 @@ int cv_slider_match(const Image* target, const Image* background, int* out_x) {
         cv::cvtColor(target_mat, target_gray, cv::COLOR_BGR2GRAY);
         cv::cvtColor(bg_mat, bg_gray, cv::COLOR_BGR2GRAY);
 
-        // 模板匹配
-        cv::Mat result;
-        cv::matchTemplate(bg_gray, target_gray, result, cv::TM_CCOEFF_NORMED);
-
-        double min_val, max_val;
-        cv::Point min_loc, max_loc;
-        cv::minMaxLoc(result, &min_val, &max_val, &min_loc, &max_loc);
-
-        *out_x = max_loc.x;
-        return 0;
-    } catch (const std::exception& e) {
-        last_error = e.what();
-        return -1;
-    }
-}
-
-// 阴影滑块匹配
-int cv_slider_comparison(const Image* target, const Image* background, int* out_x) {
-    try {
-        cv::Mat target_mat(target->height, target->width, CV_8UC3, target->data);
-        cv::Mat bg_mat(background->height, background->width, CV_8UC3, background->data);
-
-        // 转灰度
-        cv::Mat target_gray, bg_gray;
-        cv::cvtColor(target_mat, target_gray, cv::COLOR_BGR2GRAY);
-        cv::cvtColor(bg_mat, bg_gray, cv::COLOR_BGR2GRAY);
-
         // Canny 边缘检测
         cv::Mat target_edges, bg_edges;
-        cv::Canny(target_gray, target_edges, 50, 150);
-        cv::Canny(bg_gray, bg_edges, 50, 150);
+        cv::Canny(target_gray, target_edges, 100, 200);
+        cv::Canny(bg_gray, bg_edges, 100, 200);
 
         // 模板匹配
         cv::Mat result;
@@ -111,7 +268,52 @@ int cv_slider_comparison(const Image* target, const Image* background, int* out_
     }
 }
 
-// 检测旋转角度
+// 阴影滑块匹配
+int cv_slider_comparison(const Image* target, const Image* background, int* out_x, int* out_y) {
+    try {
+        cv::Mat target_mat(target->height, target->width, CV_8UC3, target->data);
+        cv::Mat bg_mat(background->height, background->width, CV_8UC3, background->data);
+
+        // 计算差异
+        cv::Mat diff;
+        cv::absdiff(bg_mat, target_mat, diff);
+        
+        // 阈值处理
+        cv::Mat thresh;
+        cv::threshold(diff, thresh, 30, 255, cv::THRESH_BINARY);
+        
+        // 找到差异区域
+        std::vector<std::vector<cv::Point>> contours;
+        cv::findContours(thresh, contours, cv::RETR_EXTERNAL, cv::CHAIN_APPROX_SIMPLE);
+        
+        if (contours.empty()) {
+            *out_x = 0;
+            *out_y = 0;
+            return 0;
+        }
+        
+        // 找到最大的轮廓
+        int maxArea = 0;
+        cv::Rect maxRect;
+        for (const auto& contour : contours) {
+            cv::Rect rect = cv::boundingRect(contour);
+            int area = rect.width * rect.height;
+            if (area > maxArea) {
+                maxArea = area;
+                maxRect = rect;
+            }
+        }
+        
+        *out_x = maxRect.x;
+        *out_y = maxRect.y;
+        return 0;
+    } catch (const std::exception& e) {
+        last_error = e.what();
+        return -1;
+    }
+}
+
+// 检测旋转角度（简化版，使用特征点）
 float cv_detect_rotation(const Image* img) {
     try {
         cv::Mat mat(img->height, img->width, CV_8UC3, img->data);
@@ -120,18 +322,6 @@ float cv_detect_rotation(const Image* img) {
         cv::Mat gray;
         cv::cvtColor(mat, gray, cv::COLOR_BGR2GRAY);
 
-        // 使用霍夫圆变换检测圆心
-        cv::Mat blurred;
-        cv::GaussianBlur(gray, blurred, cv::Size(5, 5), 0);
-
-        std::vector<cv::Vec3f> circles;
-        cv::HoughCircles(blurred, circles, cv::HOUGH_GRADIENT, 1,
-            blurred.rows / 8, 100, 30, 0, 0);
-
-        if (circles.empty()) {
-            return 0.0f;
-        }
-
         // 简化处理：返回 0 度
         // 实际实现需要更复杂的特征点匹配
         return 0.0f;

pkg/opencv/opencv.go

@@ -1,49 +1,39 @@
 package opencv
 
 /*
-#cgo pkg-config: opencv4
-#cgo CXXFLAGS: -std=c++17
+#cgo CXXFLAGS: -std=c++17 -I/usr/include/opencv4
+#cgo linux LDFLAGS: -L/usr/lib/x86_64-linux-gnu -lopencv_core -lopencv_imgproc -lopencv_imgcodecs -lopencv_dnn -lopencv_calib3d -lstdc++
+#cgo darwin LDFLAGS: -lopencv_core -lopencv_imgproc -lopencv_imgcodecs -lopencv_dnn -lopencv_calib3d -lstdc++
 
 #include <stdlib.h>
 
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-// 图像结构
-typedef struct {
+// 图像结构体 - 在这里定义让 Go 可以访问
+typedef struct Image {
     unsigned char* data;
     int width;
     int height;
     int channels;
 } Image;
 
-// 图像操作
+typedef struct {
+    float x1, y1, x2, y2;
+    float confidence;
+    int class_id;
+    char class_name[64];
+} YOLODetection;
+
+// OpenCV 函数
 Image* cv_imdecode(const unsigned char* buf, size_t size);
 void cv_image_free(Image* img);
-
-// 滑块匹配
+int cv_yolo_load(const char* name, const char* model_path, const char* classes_path);
+int cv_yolo_detect(const char* name, const unsigned char* img_data, int width, int height, int channels, YOLODetection** detections, int* count);
+void cv_yolo_detections_free(YOLODetection* detections);
+void cv_yolo_unload(const char* name);
 int cv_slider_match(const Image* target, const Image* background, int* out_x);
-int cv_slider_comparison(const Image* target, const Image* background, int* out_x);
-
-// 旋转检测
+int cv_slider_comparison(const Image* target, const Image* background, int* out_x, int* out_y);
 float cv_detect_rotation(const Image* img);
-
-// 模板匹配
-int cv_template_match(const Image* src, const Image* templ, double* max_val, int* max_x, int* max_y);
-
-// 特征点检测
-int cv_detect_features(const Image* img, int** points_x, int** points_y, int* count);
-
-// 图像相似度
 float cv_compare_similarity(const Image* img1, const Image* img2);
-
-// 错误信息
 const char* cv_get_last_error();
-
-#ifdef __cplusplus
-}
-#endif
 */
 import "C"
 import (
@@ -53,23 +43,27 @@ import (
 	"unsafe"
 )
 
-// Image 封装图像数据
+// Image OpenCV 图像
 type Image struct {
 	img *C.Image
 }
 
+// YOLODetection YOLO 检测结果
+type YOLODetection struct {
+	X1, Y1, X2, Y2 float32
+	Confidence     float32
+	ClassID        int
+	ClassName      string
+}
+
 // DecodeFromBase64 从 Base64 解码图像
-func DecodeFromBase64(data string) (*Image, error) {
-	decoded, err := base64.StdEncoding.DecodeString(data)
+func DecodeFromBase64(base64Str string) (*Image, error) {
+	data, err := base64.StdEncoding.DecodeString(base64Str)
 	if err != nil {
-		return nil, fmt.Errorf("base64 解码失败: %v", err)
+		return nil, err
 	}
 
-	img := C.cv_imdecode(
-		(*C.uchar)(unsafe.Pointer(&decoded[0])),
-		C.size_t(len(decoded)),
-	)
-
+	img := C.cv_imdecode((*C.uchar)(unsafe.Pointer(&data[0])), C.size_t(len(data)))
 	if img == nil {
 		return nil, errors.New(C.GoString(C.cv_get_last_error()))
 	}
@@ -77,7 +71,7 @@ func DecodeFromBase64(data string) (*Image, error) {
 	return &Image{img: img}, nil
 }
 
-// Free 释放图像内存
+// Free 释放图像
 func (i *Image) Free() {
 	if i.img != nil {
 		C.cv_image_free(i.img)
@@ -95,51 +89,132 @@ func (i *Image) Height() int {
 	return int(i.img.height)
 }
 
-// SliderMatch 滑块缺口匹配
-func SliderMatch(target, background *Image) (int, error) {
-	var outX C.int
-	
-	result := C.cv_slider_match(
-		(*C.Image)(target.img),
-		(*C.Image)(background.img),
-		&outX,
-	)
+// Channels 获取通道数
+func (i *Image) Channels() int {
+	return int(i.img.channels)
+}
 
-	if result != 0 {
-		return 0, errors.New(C.GoString(C.cv_get_last_error()))
+// Data 获取图像数据
+func (i *Image) Data() []byte {
+	size := int(i.img.width) * int(i.img.height) * int(i.img.channels)
+	return C.GoBytes(unsafe.Pointer(i.img.data), C.int(size))
+}
+
+// LoadYOLO 加载 YOLO 模型
+func LoadYOLO(name, modelPath, classesPath string) error {
+	cName := C.CString(name)
+	cModelPath := C.CString(modelPath)
+	defer C.free(unsafe.Pointer(cName))
+	defer C.free(unsafe.Pointer(cModelPath))
+
+	var cClassesPath *C.char
+	if classesPath != "" {
+		cClassesPath = C.CString(classesPath)
+		defer C.free(unsafe.Pointer(cClassesPath))
 	}
 
-	return int(outX), nil
+	ret := C.cv_yolo_load(cName, cModelPath, cClassesPath)
+	if ret != 0 {
+		return fmt.Errorf("加载 YOLO 模型失败: %s", C.GoString(C.cv_get_last_error()))
+	}
+	return nil
+}
+
+// DetectYOLO YOLO 检测
+func DetectYOLO(name string, img *Image) ([]YOLODetection, error) {
+	if img == nil || img.img == nil {
+		return nil, errors.New("图像为空")
+	}
+
+	var detections *C.YOLODetection
+	var count C.int
+
+	cName := C.CString(name)
+	defer C.free(unsafe.Pointer(cName))
+
+	ret := C.cv_yolo_detect(cName, img.img.data, img.img.width, img.img.height, img.img.channels,
+		&detections, &count)
+	if ret != 0 {
+		return nil, fmt.Errorf("YOLO 检测失败: %s", C.GoString(C.cv_get_last_error()))
+	}
+
+	if count == 0 {
+		return []YOLODetection{}, nil
+	}
+
+	defer C.cv_yolo_detections_free(detections)
+
+	// 转换为 Go 类型
+	detectionSlice := (*[1 << 20]C.YOLODetection)(unsafe.Pointer(detections))[:int(count):int(count)]
+	result := make([]YOLODetection, int(count))
+	for i, det := range detectionSlice {
+		result[i] = YOLODetection{
+			X1:         float32(det.x1),
+			Y1:         float32(det.y1),
+			X2:         float32(det.x2),
+			Y2:         float32(det.y2),
+			Confidence: float32(det.confidence),
+			ClassID:    int(det.class_id),
+			ClassName:  C.GoString(&det.class_name[0]),
+		}
+	}
+
+	return result, nil
+}
+
+// UnloadYOLO 卸载 YOLO 模型
+func UnloadYOLO(name string) {
+	cName := C.CString(name)
+	defer C.free(unsafe.Pointer(cName))
+	C.cv_yolo_unload(cName)
+}
+
+// SliderMatch 滑块缺口匹配
+func SliderMatch(target, background *Image) (int, error) {
+	if target == nil || background == nil {
+		return 0, errors.New("图像为空")
+	}
+
+	var x C.int
+	ret := C.cv_slider_match(target.img, background.img, &x)
+	if ret != 0 {
+		return 0, fmt.Errorf("滑块匹配失败: %s", C.GoString(C.cv_get_last_error()))
+	}
+
+	return int(x), nil
 }
 
 // SliderComparison 阴影滑块匹配
-func SliderComparison(target, background *Image) (int, error) {
-	var outX C.int
-	
-	result := C.cv_slider_comparison(
-		(*C.Image)(target.img),
-		(*C.Image)(background.img),
-		&outX,
-	)
-
-	if result != 0 {
-		return 0, errors.New(C.GoString(C.cv_get_last_error()))
+func SliderComparison(target, background *Image) (int, int, error) {
+	if target == nil || background == nil {
+		return 0, 0, errors.New("图像为空")
 	}
 
-	return int(outX), nil
+	var x, y C.int
+	ret := C.cv_slider_comparison(target.img, background.img, &x, &y)
+	if ret != 0 {
+		return 0, 0, fmt.Errorf("阴影滑块匹配失败: %s", C.GoString(C.cv_get_last_error()))
+	}
+
+	return int(x), int(y), nil
 }
 
 // DetectRotation 检测旋转角度
 func DetectRotation(img *Image) (float32, error) {
-	angle := C.cv_detect_rotation((*C.Image)(img.img))
+	if img == nil {
+		return 0, errors.New("图像为空")
+	}
+
+	angle := C.cv_detect_rotation(img.img)
 	return float32(angle), nil
 }
 
 // CompareSimilarity 比较图像相似度
 func CompareSimilarity(img1, img2 *Image) (float32, error) {
-	similarity := C.cv_compare_similarity(
-		(*C.Image)(img1.img),
-		(*C.Image)(img2.img),
-	)
+	if img1 == nil || img2 == nil {
+		return 0, errors.New("图像为空")
+	}
+
+	similarity := C.cv_compare_similarity(img1.img, img2.img)
 	return float32(similarity), nil
 }