bushu

yemian/yemian_line/2line—distance.py

@@ -0,0 +1,227 @@
+# ===================================================
+# final_compare_corner.py
+# 同时显示 Canny 物理边缘（红线）和 YOLO 预测左边缘（绿线）
+# 基于角点拟合直线，剔除离群点
+# ===================================================
+
+import os
+import cv2
+import numpy as np
+from pathlib import Path
+from ultralytics import YOLO
+
+# ============================
+# 参数
+# ============================
+TARGET_SIZE = 640
+MAX_CORNERS = 200
+QUALITY_LEVEL = 0.01
+MIN_DISTANCE = 5
+DIST_THRESH = 15
+ROIS = [(859, 810, 696, 328)]  # 全局 ROI，可按需修改
+OUTPUT_DIR = "./final_output"
+
+# ============================
+# Canny 边缘部分（保持原逻辑）
+# ============================
+def load_global_rois(txt_path):
+    rois = []
+    if not os.path.exists(txt_path):
+        print(f"❌ ROI 文件不存在: {txt_path}")
+        return rois
+    with open(txt_path, 'r') as f:
+        for line in f:
+            s = line.strip()
+            if s:
+                try:
+                    x, y, w, h = map(int, s.split(','))
+                    rois.append((x, y, w, h))
+                except Exception as e:
+                    print(f"⚠️ 无法解析 ROI 行 '{s}': {e}")
+    return rois
+
+def fit_line_best(points, distance_thresh=5, max_iter=5):
+    if len(points) < 2:
+        return None
+    points = points.astype(np.float32)
+    for _ in range(max_iter):
+        mean = np.mean(points, axis=0)
+        cov = np.cov(points.T)
+        eigvals, eigvecs = np.linalg.eig(cov)
+        idx = np.argmax(eigvals)
+        direction = eigvecs[:, idx]
+        vx, vy = direction
+        x0, y0 = mean
+        dists = np.abs(vy*(points[:,0]-x0) - vx*(points[:,1]-y0)) / np.hypot(vx, vy)
+        inliers = points[dists <= distance_thresh]
+        if len(inliers) == len(points) or len(inliers) < 2:
+            break
+        points = inliers
+    if len(points) < 2:
+        return None
+    X = points[:, 0].reshape(-1, 1)
+    y = points[:, 1]
+    try:
+        from sklearn.linear_model import RANSACRegressor
+        ransac = RANSACRegressor(residual_threshold=distance_thresh)
+        ransac.fit(X, y)
+        k = ransac.estimator_.coef_[0]
+        b = ransac.estimator_.intercept_
+        vx = 1 / np.sqrt(1 + k**2)
+        vy = k / np.sqrt(1 + k**2)
+        x0 = np.mean(points[:,0])
+        y0 = k*x0 + b
+    except:
+        mean = np.mean(points, axis=0)
+        cov = np.cov(points.T)
+        eigvals, eigvecs = np.linalg.eig(cov)
+        idx = np.argmax(eigvals)
+        direction = eigvecs[:, idx]
+        vx, vy = direction
+        x0, y0 = mean
+    return vx, vy, x0, y0
+
+def extract_canny_overlay(image_path, roi_file, distance_thresh=3):
+    img = cv2.imread(image_path)
+    if img is None:
+        print(f"❌ 无法读取图片: {image_path}")
+        return None
+
+    overlay_img = img.copy()
+    rois = load_global_rois(roi_file)
+    if not rois:
+        print("❌ 没有有效 ROI")
+        return overlay_img
+
+    for idx, (x, y, w, h) in enumerate(rois):
+        roi = img[y:y+h, x:x+w]
+        gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
+        edges = cv2.Canny(gray, 100, 200)
+        contours, _ = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
+        longest_contour = max(contours, key=lambda c: cv2.arcLength(c, closed=False), default=None)
+
+        if longest_contour is not None and len(longest_contour) >= 2:
+            points = longest_contour.reshape(-1, 2)
+            line = fit_line_best(points, distance_thresh=distance_thresh, max_iter=5)
+            if line is not None:
+                vx, vy, x0, y0 = line
+                cols = w
+                lefty = int(y0 - vy/vx * x0)
+                righty = int(y0 + vy/vx * (cols - x0))
+                pt1 = (x, y + lefty)
+                pt2 = (x + cols - 1, y + righty)
+                cv2.line(overlay_img, pt1, pt2, (0, 0, 255), 2)  # 红色
+                print(f"✅ ROI {idx} Canny 边缘拟合完成")
+    return overlay_img
+
+# ============================
+# YOLO 角点 + 拟合直线
+# ============================
+def select_edge_corners(corners, w, left_ratio=0.2, right_ratio=0.2, y_var_thresh=5):
+    if corners is None:
+        return np.zeros((0,2), dtype=np.int32), np.zeros((0,2), dtype=np.int32)
+    corners = np.int32(corners).reshape(-1,2)
+    left_thresh = int(w*left_ratio)
+    right_thresh = w - int(w*right_ratio)
+    left_candidates = corners[corners[:,0]<=left_thresh]
+    right_candidates = corners[corners[:,0]>=right_thresh]
+
+    def filter_by_y_variation(pts):
+        if len(pts)<2:
+            return pts.astype(np.int32)
+        pts_sorted = pts[np.argsort(pts[:,1])]
+        diffs = np.abs(np.diff(pts_sorted[:,1]))
+        keep_idx = np.where(diffs>y_var_thresh)[0]
+        if len(keep_idx)==0:
+            return pts_sorted.astype(np.int32)
+        selected = [pts_sorted[i] for i in keep_idx]+[pts_sorted[i+1] for i in keep_idx]
+        selected = np.array(selected)
+        _, idx = np.unique(selected.reshape(-1,2), axis=0, return_index=True)
+        selected = selected[np.sort(idx)]
+        return selected.astype(np.int32)
+
+    return filter_by_y_variation(left_candidates), filter_by_y_variation(right_candidates)
+
+def fit_line_with_outlier_removal(pts, dist_thresh=DIST_THRESH):
+    if pts is None or len(pts)<2:
+        return None, np.zeros((0,2), dtype=np.int32)
+    pts = np.array(pts, dtype=np.float64)
+    x, y = pts[:,0], pts[:,1]
+    try:
+        m, b = np.polyfit(y, x, 1)
+    except:
+        return None, pts.astype(np.int32)
+    x_fit = m*y + b
+    mask = np.abs(x-x_fit)<dist_thresh
+    if np.sum(mask)<2:
+        return (m,b), pts.astype(np.int32)
+    x2, y2 = x[mask], y[mask]
+    m2, b2 = np.polyfit(y2, x2, 1)
+    inliers = np.stack([x2,y2],axis=1).astype(np.int32)
+    return (m2,b2), inliers
+
+def get_yolo_left_edge_lines(image_path, model_path, rois=ROIS, imgsz=TARGET_SIZE):
+    model = YOLO(model_path)
+    img = cv2.imread(image_path)
+    if img is None:
+        print(f"❌ 无法读取图片: {image_path}")
+        return []
+    lines = []
+
+    for (x, y, w, h) in rois:
+        roi_img = img[y:y+h, x:x+w]
+        resized = cv2.resize(roi_img, (imgsz, imgsz))
+        results = model(resized, imgsz=imgsz, verbose=False)
+        result = results[0]
+        if result.masks is None or len(result.masks.data)==0:
+            continue
+
+        mask = result.masks.data[0].cpu().numpy()
+        mask_bin = (mask>0.5).astype(np.uint8)
+        mask_bin = cv2.resize(mask_bin, (w,h), interpolation=cv2.INTER_NEAREST)
+
+        # 角点检测
+        mask_gray = (mask_bin*255).astype(np.uint8)
+        corners = cv2.goodFeaturesToTrack(mask_gray, maxCorners=MAX_CORNERS,
+                                          qualityLevel=QUALITY_LEVEL, minDistance=MIN_DISTANCE)
+        left_pts, _ = select_edge_corners(corners, w)
+        line_params, inliers = fit_line_with_outlier_removal(left_pts)
+        if line_params is None:
+            continue
+        m,b = line_params
+        y1, y2 = 0, h-1
+        x1 = int(m*y1 + b)
+        x2 = int(m*y2 + b)
+        lines.append((x+x1, y+y1, x+x2, y+y2))
+    return lines
+
+# ============================
+# 对比融合
+# ============================
+def compare_canny_vs_yolo(image_path, canny_roi_file, model_path, output_dir=OUTPUT_DIR):
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    canny_img = extract_canny_overlay(image_path, canny_roi_file, distance_thresh=6)
+    if canny_img is None:
+        return
+
+    yolo_lines = get_yolo_left_edge_lines(image_path, model_path)
+    result_img = canny_img.copy()
+    for x1,y1,x2,y2 in yolo_lines:
+        cv2.line(result_img, (x1,y1), (x2,y2), (0,255,0), 2) # 绿色
+        cv2.circle(result_img, (x1,y1), 4, (255,0,0), -1)   # 蓝色起点
+
+    output_path = output_dir / f"compare_{Path(image_path).stem}.jpg"
+    cv2.imwrite(str(output_path), result_img)
+    print(f"✅ 对比图已保存: {output_path}")
+
+# ============================
+# 使用示例
+# ============================
+if __name__ == "__main__":
+    IMAGE_PATH = "../test_image/2.jpg"
+    CANNY_ROI_FILE = "../roi_coordinates/1_rois1.txt"
+    MODEL_PATH = "best.pt"
+
+    compare_canny_vs_yolo(IMAGE_PATH, CANNY_ROI_FILE, MODEL_PATH)

yemian/yemian_line/gaiban_edge.py

@@ -0,0 +1,149 @@
+import os
+import cv2
+import numpy as np
+from pathlib import Path
+from sklearn.linear_model import RANSACRegressor
+
+# ---------------------------
+# 读取 ROI 列表 txt
+# ---------------------------
+def load_rois_from_txt(txt_path):
+    rois = []
+    if not os.path.exists(txt_path):
+        print(f"❌ ROI 文件不存在: {txt_path}")
+        return rois
+    with open(txt_path, 'r') as f:
+        for line in f:
+            s = line.strip()
+            if s:
+                try:
+                    x, y, w, h = map(int, s.split(','))
+                    rois.append((x, y, w, h))
+                except Exception as e:
+                    print(f"⚠️ 无法解析 ROI 行 '{s}': {e}")
+    return rois
+
+# ---------------------------
+# PCA + RANSAC + 迭代去离群点拟合直线
+# ---------------------------
+def fit_line_best(points, distance_thresh=5, max_iter=5):
+    if len(points) < 2:
+        return None
+
+    points = points.astype(np.float32)
+
+    for _ in range(max_iter):
+        mean = np.mean(points, axis=0)
+        cov = np.cov(points.T)
+        eigvals, eigvecs = np.linalg.eig(cov)
+        idx = np.argmax(eigvals)
+        direction = eigvecs[:, idx]
+        vx, vy = direction
+        x0, y0 = mean
+
+        dists = np.abs(vy*(points[:,0]-x0) - vx*(points[:,1]-y0)) / np.hypot(vx, vy)
+        inliers = points[dists <= distance_thresh]
+
+        if len(inliers) == len(points) or len(inliers) < 2:
+            break
+        points = inliers
+
+    if len(points) < 2:
+        return None
+
+    # RANSAC 拟合 y = kx + b
+    X = points[:, 0].reshape(-1, 1)
+    y = points[:, 1]
+    try:
+        ransac = RANSACRegressor(residual_threshold=distance_thresh)
+        ransac.fit(X, y)
+        k = ransac.estimator_.coef_[0]
+        b = ransac.estimator_.intercept_
+
+        vx = 1 / np.sqrt(1 + k**2)
+        vy = k / np.sqrt(1 + k**2)
+        x0 = np.mean(points[:,0])
+        y0 = k*x0 + b
+    except:
+        mean = np.mean(points, axis=0)
+        cov = np.cov(points.T)
+        eigvals, eigvecs = np.linalg.eig(cov)
+        idx = np.argmax(eigvals)
+        direction = eigvecs[:, idx]
+        vx, vy = direction
+        x0, y0 = mean
+
+    return vx, vy, x0, y0
+
+# ---------------------------
+# 封装函数：读取 ROI txt -> 拟合直线 -> 可视化
+# ---------------------------
+def fit_lines_from_image_txt(image_path, roi_txt_path, distance_thresh=5, draw_overlay=True):
+    """
+    输入：
+        image_path: 原图路径
+        roi_txt_path: ROI txt 文件路径，每行 x,y,w,h
+        distance_thresh: 直线拟合残差阈值
+        draw_overlay: 是否在原图上叠加拟合直线
+    输出：
+        lines: 每个 ROI 的拟合直线 [(vx, vy, x0, y0), ...]
+        overlay_img: 可视化原图叠加拟合直线
+    """
+    img = cv2.imread(image_path)
+    if img is None:
+        print(f"❌ 无法读取图片: {image_path}")
+        return [], None
+
+    rois = load_rois_from_txt(roi_txt_path)
+    if not rois:
+        print("❌ 没有有效 ROI")
+        return [], None
+
+    overlay_img = img.copy() if draw_overlay else None
+    lines = []
+
+    for idx, (x, y, w, h) in enumerate(rois):
+        roi = img[y:y+h, x:x+w]
+        gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
+        edges = cv2.Canny(gray, 100, 200)
+        contours, _ = cv2.findContours(edges, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
+
+        longest_contour = max(contours, key=lambda c: cv2.arcLength(c, closed=False), default=None)
+
+        if longest_contour is not None and len(longest_contour) >= 2:
+            points = longest_contour.reshape(-1, 2)
+            line = fit_line_best(points, distance_thresh=distance_thresh, max_iter=5)
+            if line:
+                vx, vy, x0, y0 = line
+                lines.append(line)
+
+                if draw_overlay:
+                    cols = gray.shape[1]
+                    lefty = int(y0 - vy/vx * x0)
+                    righty = int(y0 + vy/vx * (cols - x0))
+
+                    # 绘制在原图
+                    cv2.line(overlay_img, (x, y + lefty), (x + cols - 1, y + righty), (0, 0, 255), 2)
+                    cv2.drawContours(overlay_img, [longest_contour + np.array([x, y])], -1, (0, 255, 0), 1)
+            else:
+                lines.append(None)
+        else:
+            lines.append(None)
+
+    return lines, overlay_img
+
+# ---------------------------
+# 使用示例
+# ---------------------------
+if __name__ == "__main__":
+    image_path = "../test_image/1.jpg"
+    roi_txt_path = "../roi_coordinates/1_rois1.txt"
+
+    lines, overlay = fit_lines_from_image_txt(image_path, roi_txt_path, distance_thresh=5)
+
+    for idx, line in enumerate(lines):
+        print(f"ROI {idx} 拟合直线: {line}")
+
+    if overlay is not None:
+        cv2.imwrite("overlay_result.jpg", overlay)
+        print("✅ 原图叠加拟合直线已保存: overlay_result.jpg")

yemian/yemian_line/main—tuili_line.py

@@ -0,0 +1,161 @@
+import os
+import cv2
+import numpy as np
+from pathlib import Path
+from ultralytics import YOLO
+
+TARGET_SIZE = 640  # 模型输入尺寸
+
+# --------------------
+# 全局 ROI 定义
+# --------------------
+ROIS = [
+    (859, 810, 696, 328),  # (x, y, w, h)
+]
+
+# --------------------
+# 根据角点分布，选取左右边缘角点
+# --------------------
+def select_edge_corners(corners, w, left_ratio=0.2, right_ratio=0.2, y_var_thresh=5):
+    if corners is None:
+        return [], []
+
+    corners = np.int32(corners).reshape(-1, 2)
+    x_min, x_max = 0, w
+
+    left_thresh = x_min + int(w * left_ratio)
+    right_thresh = x_max - int(w * right_ratio)
+
+    # 左右候选角点
+    left_candidates = corners[corners[:, 0] <= left_thresh]
+    right_candidates = corners[corners[:, 0] >= right_thresh]
+
+    # --------------------
+    # 进一步按 y 变化筛选
+    # --------------------
+    def filter_by_y_variation(pts):
+        if len(pts) < 2:
+            return pts
+        pts_sorted = pts[np.argsort(pts[:, 1])]
+        diffs = np.abs(np.diff(pts_sorted[:, 1]))
+        keep_idx = np.where(diffs > y_var_thresh)[0]
+        selected = [pts_sorted[i] for i in keep_idx] + [pts_sorted[i + 1] for i in keep_idx]
+        return np.array(selected) if len(selected) > 0 else pts_sorted
+
+    left_final = filter_by_y_variation(left_candidates)
+    right_final = filter_by_y_variation(right_candidates)
+
+    return left_final, right_final
+
+
+# --------------------
+# 拟合直线并剔除离散点
+# --------------------
+def fit_line_with_outlier_removal(pts, dist_thresh=15):
+    """
+    pts: (N,2) array
+    dist_thresh: 点到拟合直线的最大允许距离
+    返回 (m, b) 直线参数, 以及拟合用到的点
+    """
+    if pts is None or len(pts) < 2:
+        return None, pts
+
+    pts = np.array(pts)
+    x, y = pts[:, 0], pts[:, 1]
+
+    # 第一次拟合
+    m, b = np.polyfit(y, x, 1)  # x = m*y + b
+    x_fit = m * y + b
+    dists = np.abs(x - x_fit)
+
+    # 剔除离群点
+    mask = dists < dist_thresh
+    x2, y2 = x[mask], y[mask]
+
+    if len(x2) < 2:
+        return (m, b), pts  # 保底返回
+
+    # 二次拟合
+    m, b = np.polyfit(y2, x2, 1)
+    return (m, b), np.stack([x2, y2], axis=1)
+
+
+# --------------------
+# 推理 ROI 并可视化 mask + 边缘角点 + 拟合直线
+# --------------------
+def infer_mask_with_selected_corners(image_path, model_path, output_dir="./output"):
+    model = YOLO(model_path)
+
+    image_path = Path(image_path)
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    orig_img = cv2.imread(str(image_path))
+    overlay_img = orig_img.copy()
+
+    for idx, (x, y, w, h) in enumerate(ROIS):
+        roi_img = orig_img[y:y+h, x:x+w]
+        resized_img = cv2.resize(roi_img, (TARGET_SIZE, TARGET_SIZE))
+
+        # 模型推理
+        results = model(source=resized_img, imgsz=TARGET_SIZE, verbose=False)
+        result = results[0]
+
+        # 可视化 mask
+        if result.masks is not None and len(result.masks.data) > 0:
+            mask = result.masks.data[0].cpu().numpy()
+            mask_bin = (mask > 0.5).astype(np.uint8)
+            mask_bin = cv2.resize(mask_bin, (w, h), interpolation=cv2.INTER_NEAREST)
+
+            # 绿色 mask 覆盖
+            color_mask = np.zeros_like(roi_img, dtype=np.uint8)
+            color_mask[mask_bin == 1] = (0, 255, 0)
+            overlay_img[y:y+h, x:x+w] = cv2.addWeighted(roi_img, 0.7, color_mask, 0.3, 0)
+
+            # 角点检测
+            mask_gray = (mask_bin * 255).astype(np.uint8)
+            corners = cv2.goodFeaturesToTrack(mask_gray,
+                                              maxCorners=200,
+                                              qualityLevel=0.01,
+                                              minDistance=5)
+
+            # 选择左右边缘角点
+            left_pts, right_pts = select_edge_corners(corners, w)
+
+            # 拟合直线并剔除离散点
+            left_line, left_inliers = fit_line_with_outlier_removal(left_pts)
+            right_line, right_inliers = fit_line_with_outlier_removal(right_pts)
+
+            # 可视化角点
+            for cx, cy in left_inliers:
+                cv2.circle(overlay_img[y:y+h, x:x+w], (int(cx), int(cy)), 5, (0, 0, 255), -1)
+            for cx, cy in right_inliers:
+                cv2.circle(overlay_img[y:y+h, x:x+w], (int(cx), int(cy)), 5, (255, 0, 0), -1)
+
+            # 可视化拟合直线
+            if left_line is not None:
+                m, b = left_line
+                y1, y2 = 0, h
+                x1, x2 = int(m * y1 + b), int(m * y2 + b)
+                cv2.line(overlay_img[y:y+h, x:x+w], (x1, y1), (x2, y2), (0, 0, 255), 3)
+
+            if right_line is not None:
+                m, b = right_line
+                y1, y2 = 0, h
+                x1, x2 = int(m * y1 + b), int(m * y2 + b)
+                cv2.line(overlay_img[y:y+h, x:x+w], (x1, y1), (x2, y2), (255, 0, 0), 3)
+
+    # 保存结果
+    save_path = output_dir / f"mask_edge_corners_{image_path.name}"
+    cv2.imwrite(str(save_path), overlay_img)
+    print(f"✅ 保存结果: {save_path}")
+
+    return overlay_img
+
+
+# ===================== 使用示例 =====================
+if __name__ == "__main__":
+    IMAGE_PATH = "../test_image/1.jpg"
+    MODEL_PATH = "best.pt"
+
+    infer_mask_with_selected_corners(IMAGE_PATH, MODEL_PATH)

yemian/yemian_line/tuili_c_f.py

@@ -0,0 +1,190 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+import os
+import cv2
+import numpy as np
+from pathlib import Path
+from ultralytics import YOLO
+
+# --------------------
+# 参数设置（固定在脚本中）
+# --------------------
+INPUT_DIR = "/media/hx/04e879fa-d697-4b02-ac7e-a4148876ebb0/dataset/test_l"        # 图片文件夹
+MODEL_PATH = "best.pt"             # YOLO 模型
+OUTPUT_DIR = "./output"            # 保存结果
+TARGET_SIZE = 640                  # YOLO 输入尺寸
+DIST_THRESH = 15                   # 剔除离群点阈值
+MAX_CORNERS = 200                  # goodFeaturesToTrack 最大角点数
+QUALITY_LEVEL = 0.01               # goodFeaturesToTrack qualityLevel
+MIN_DISTANCE = 5                   # goodFeaturesToTrack minDistance
+
+# 全局 ROI 定义
+ROIS = [
+    (859, 810, 696, 328),  # (x, y, w, h)
+]
+
+# --------------------
+# 左右边缘角点筛选
+# --------------------
+def select_edge_corners(corners, w, left_ratio=0.2, right_ratio=0.2, y_var_thresh=5):
+    if corners is None:
+        return np.zeros((0,2), dtype=np.int32), np.zeros((0,2), dtype=np.int32)
+
+    corners = np.int32(corners).reshape(-1, 2)
+    x_min, x_max = 0, w
+    left_thresh = x_min + int(w * left_ratio)
+    right_thresh = x_max - int(w * right_ratio)
+
+    left_candidates = corners[corners[:,0] <= left_thresh]
+    right_candidates = corners[corners[:,0] >= right_thresh]
+
+    def filter_by_y_variation(pts):
+        if len(pts) < 2:
+            return pts.astype(np.int32)
+        pts_sorted = pts[np.argsort(pts[:,1])]
+        diffs = np.abs(np.diff(pts_sorted[:,1]))
+        keep_idx = np.where(diffs > y_var_thresh)[0]
+        if len(keep_idx) == 0:
+            return pts_sorted.astype(np.int32)
+        selected = [pts_sorted[i] for i in keep_idx] + [pts_sorted[i+1] for i in keep_idx]
+        selected = np.array(selected)
+        selected = selected[np.argsort(selected[:,1])]
+        _, idx = np.unique(selected.reshape(-1,2), axis=0, return_index=True)
+        selected = selected[np.sort(idx)]
+        return selected.astype(np.int32)
+
+    left_final = filter_by_y_variation(left_candidates)
+    right_final = filter_by_y_variation(right_candidates)
+
+    return left_final, right_final
+
+# --------------------
+# 拟合直线并剔除离散点
+# --------------------
+def fit_line_with_outlier_removal(pts, dist_thresh=DIST_THRESH):
+    if pts is None or len(pts) < 2:
+        return None, np.zeros((0,2), dtype=np.int32)
+
+    pts = np.array(pts, dtype=np.float64)
+    x = pts[:,0]
+    y = pts[:,1]
+
+    try:
+        m, b = np.polyfit(y, x, 1)
+    except:
+        return None, np.zeros((0,2), dtype=np.int32)
+
+    x_fit = m*y + b
+    dists = np.abs(x - x_fit)
+    mask = dists < dist_thresh
+
+    if np.sum(mask) < 2:
+        return (m,b), pts.astype(np.int32)
+
+    x2, y2 = x[mask], y[mask]
+    try:
+        m2, b2 = np.polyfit(y2, x2, 1)
+    except:
+        return (m,b), np.stack([x2,y2],axis=1).astype(np.int32)
+
+    inliers = np.stack([x2,y2],axis=1).astype(np.int32)
+    return (m2,b2), inliers
+
+# --------------------
+# 单张图 ROI 处理
+# --------------------
+def process_roi_on_image(orig_img, roi):
+    rx, ry, rw, rh = roi
+    h_img, w_img = orig_img.shape[:2]
+    rx = max(0, rx); ry = max(0, ry)
+    rw = min(rw, w_img - rx); rh = min(rh, h_img - ry)
+    roi_img = orig_img[ry:ry+rh, rx:rx+rw].copy()
+    if roi_img.size == 0:
+        return None
+
+    resized = cv2.resize(roi_img, (TARGET_SIZE, TARGET_SIZE))
+    results = MODEL(resized, imgsz=TARGET_SIZE, verbose=False)
+    result = results[0]
+
+    overlay_roi = roi_img.copy()
+    if result.masks is None or len(result.masks.data)==0:
+        return overlay_roi
+
+    mask = result.masks.data[0].cpu().numpy()
+    mask_bin = (mask>0.5).astype(np.uint8)
+    mask_bin = cv2.resize(mask_bin,(rw,rh), interpolation=cv2.INTER_NEAREST)
+
+    # mask 半透明覆盖
+    color_mask = np.zeros_like(overlay_roi, dtype=np.uint8)
+    color_mask[mask_bin==1] = (0,255,0)
+    overlay_roi = cv2.addWeighted(overlay_roi,0.7,color_mask,0.3,0)
+
+    # 角点检测
+    mask_gray = (mask_bin*255).astype(np.uint8)
+    corners = cv2.goodFeaturesToTrack(mask_gray, maxCorners=MAX_CORNERS,
+                                      qualityLevel=QUALITY_LEVEL, minDistance=MIN_DISTANCE)
+    left_pts, right_pts = select_edge_corners(corners, rw)
+    left_line, left_inliers = fit_line_with_outlier_removal(left_pts)
+    right_line, right_inliers = fit_line_with_outlier_removal(right_pts)
+
+    # 可视化 inliers
+    for (cx,cy) in left_inliers:
+        cv2.circle(overlay_roi,(int(cx),int(cy)),4,(0,0,255),-1)
+    for (cx,cy) in right_inliers:
+        cv2.circle(overlay_roi,(int(cx),int(cy)),4,(255,0,0),-1)
+
+    # 拟合直线
+    if left_line is not None:
+        m,b = left_line
+        y1,y2 = 0, rh-1
+        x1 = int(m*y1+b); x2 = int(m*y2+b)
+        cv2.line(overlay_roi,(x1,y1),(x2,y2),(0,0,200),3)
+    if right_line is not None:
+        m,b = right_line
+        y1,y2 = 0, rh-1
+        x1 = int(m*y1+b); x2 = int(m*y2+b)
+        cv2.line(overlay_roi,(x1,y1),(x2,y2),(200,0,0),3)
+
+    return overlay_roi
+
+# --------------------
+# 批量推理文件夹
+# --------------------
+def infer_folder_images():
+    input_dir = Path(INPUT_DIR)
+    output_dir = Path(OUTPUT_DIR)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    exts = ('*.jpg','*.jpeg','*.png','*.bmp','*.tif','*.tiff')
+    files = []
+    for e in exts:
+        files.extend(sorted(input_dir.glob(e)))
+    if len(files)==0:
+        print("未找到图片文件")
+        return
+    print(f"找到 {len(files)} 张图片，开始推理...")
+
+    for img_path in files:
+        print("-> 处理:", img_path.name)
+        orig_img = cv2.imread(str(img_path))
+        if orig_img is None:
+            print("  无法读取，跳过")
+            continue
+        out_img = orig_img.copy()
+        for roi in ROIS:
+            overlay_roi = process_roi_on_image(orig_img, roi)
+            if overlay_roi is not None:
+                rx,ry,rw,rh = roi
+                h,w = overlay_roi.shape[:2]
+                out_img[ry:ry+h, rx:rx+w] = overlay_roi
+        save_path = output_dir / f"mask_edge_corners_{img_path.name}"
+        cv2.imwrite(str(save_path), out_img)
+        print("  已保存 ->", save_path.name)
+    print("批量推理完成，结果保存在:", output_dir)
+
+# --------------------
+# 主程序
+# --------------------
+if __name__ == "__main__":
+    MODEL = YOLO(MODEL_PATH)
+    infer_folder_images()

yemian/yemian_line/tuili_corner.py

@@ -0,0 +1,130 @@
+import os
+import cv2
+import numpy as np
+from pathlib import Path
+from ultralytics import YOLO
+
+TARGET_SIZE = 640  # 模型输入尺寸
+
+# --------------------
+# 全局 ROI 定义
+# --------------------
+ROIS = [
+    (859,810,696,328),  # (x, y, w, h)
+]
+
+# --------------------
+# 剔除相邻 x 差距过大的离散点
+# --------------------
+def filter_outliers_by_x(pts, x_thresh=30):
+    if len(pts) < 2:
+        return pts
+    pts_sorted = pts[np.argsort(pts[:, 1])]  # 按 y 排序
+    clean_pts = [pts_sorted[0]]
+    for i in range(1, len(pts_sorted)):
+        if abs(pts_sorted[i, 0] - pts_sorted[i-1, 0]) < x_thresh:
+            clean_pts.append(pts_sorted[i])
+    return np.array(clean_pts, dtype=np.int32)
+
+# --------------------
+# 根据角点分布，选取左右边缘角点
+# --------------------
+def select_edge_corners(corners, w, left_ratio=0.2, right_ratio=0.2, y_var_thresh=5, x_var_thresh=30):
+    if corners is None:
+        return [], []
+
+    corners = np.int32(corners).reshape(-1, 2)
+    x_min, x_max = 0, w
+
+    left_thresh = x_min + int(w * left_ratio)
+    right_thresh = x_max - int(w * right_ratio)
+
+    # 左右候选角点
+    left_candidates = corners[corners[:, 0] <= left_thresh]
+    right_candidates = corners[corners[:, 0] >= right_thresh]
+
+    # --------------------
+    # 进一步按 y 变化筛选
+    # --------------------
+    def filter_by_y_variation(pts):
+        if len(pts) < 2:
+            return pts
+        pts_sorted = pts[np.argsort(pts[:, 1])]
+        diffs = np.abs(np.diff(pts_sorted[:, 1]))
+        keep_idx = np.where(diffs > y_var_thresh)[0]
+        selected = [pts_sorted[i] for i in keep_idx] + [pts_sorted[i+1] for i in keep_idx]
+        return np.array(selected) if len(selected) > 0 else pts_sorted
+
+    left_final = filter_by_y_variation(left_candidates)
+    right_final = filter_by_y_variation(right_candidates)
+
+    # --------------------
+    # 再剔除相邻 x 值差距过大的离散点
+    # --------------------
+    left_final = filter_outliers_by_x(left_final, x_var_thresh)
+    right_final = filter_outliers_by_x(right_final, x_var_thresh)
+
+    return left_final, right_final
+
+# --------------------
+# 推理 ROI 并可视化 mask + 边缘角点
+# --------------------
+def infer_mask_with_selected_corners(image_path, model_path, output_dir="./output"):
+    model = YOLO(model_path)
+
+    image_path = Path(image_path)
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    orig_img = cv2.imread(str(image_path))
+    overlay_img = orig_img.copy()
+
+    for idx, (x, y, w, h) in enumerate(ROIS):
+        roi_img = orig_img[y:y+h, x:x+w]
+        resized_img = cv2.resize(roi_img, (TARGET_SIZE, TARGET_SIZE))
+
+        # 模型推理
+        results = model(source=resized_img, imgsz=TARGET_SIZE, verbose=False)
+        result = results[0]
+
+        # 可视化 mask
+        if result.masks is not None and len(result.masks.data) > 0:
+            mask = result.masks.data[0].cpu().numpy()
+            mask_bin = (mask > 0.5).astype(np.uint8)
+            mask_bin = cv2.resize(mask_bin, (w, h), interpolation=cv2.INTER_NEAREST)
+
+            # 绿色 mask 覆盖
+            color_mask = np.zeros_like(roi_img, dtype=np.uint8)
+            color_mask[mask_bin == 1] = (0, 255, 0)
+            overlay_img[y:y+h, x:x+w] = cv2.addWeighted(roi_img, 0.7, color_mask, 0.3, 0)
+
+            # 角点检测
+            mask_gray = (mask_bin * 255).astype(np.uint8)
+            corners = cv2.goodFeaturesToTrack(mask_gray,
+                                              maxCorners=200,
+                                              qualityLevel=0.01,
+                                              minDistance=5)
+
+            # 选择左右边缘角点
+            left_pts, right_pts = select_edge_corners(corners, w)
+
+            # 可视化
+            for cx, cy in left_pts:
+                cv2.circle(overlay_img[y:y+h, x:x+w], (cx, cy), 6, (0, 0, 255), -1)  # 左边红色
+            for cx, cy in right_pts:
+                cv2.circle(overlay_img[y:y+h, x:x+w], (cx, cy), 6, (255, 0, 0), -1)  # 右边蓝色
+
+    # 保存结果
+    save_path = output_dir / f"mask_edge_corners_{image_path.name}"
+    cv2.imwrite(str(save_path), overlay_img)
+    print(f"✅ 保存结果: {save_path}")
+
+    return overlay_img
+
+
+# ===================== 使用示例 =====================
+if __name__ == "__main__":
+    IMAGE_PATH = "../test_image/1.jpg"
+    MODEL_PATH = "best.pt"
+
+    infer_mask_with_selected_corners(IMAGE_PATH, MODEL_PATH)