Function calculate_stroke_width

modules/basic_shape_processor.py:185–263

计算边框粗细 (Stroke Width) 逻辑：沿四边向内扫描，寻找颜色突变点，多个采样点综合取中位数。 优化： - 提高突变阈值（35），减少误检 - 限制最大宽度（8像素），避免过粗 - 大多数边框在 1-5 像素

(image: np.ndarray, bbox: list, max_width: int = 8)

# ======================== 边框宽度检测 ========================
def calculate_stroke_width(image: np.ndarray, bbox: list, max_width: int = 8) -> int:
    """
    计算边框粗细 (Stroke Width)
    逻辑：沿四边向内扫描，寻找颜色突变点，多个采样点综合取中位数。
    
    优化：
    - 提高突变阈值（35），减少误检
    - 限制最大宽度（8像素），避免过粗
    - 大多数边框在 1-5 像素
    """
    x1, y1, x2, y2 = map(int, bbox)
    h, w = image.shape[:2]
    
    x1, y1 = max(0, x1), max(0, y1)
    x2, y2 = min(w, x2), min(h, y2)
    
    roi = image[y1:y2, x1:x2]
    if roi.size == 0:
        return 1
    
    roi_h, roi_w = roi.shape[:2]
    roi_gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
    
    scan_limit = min(max_width, roi_w // 2 - 1, roi_h // 2 - 1)
    if scan_limit < 1:
        return 1
    
    detected_widths = []
    
    def scan_line(pixels, limit):
        if len(pixels) < limit + 1:
            return None
        diffs = np.abs(np.diff(pixels[:limit+2].astype(int)))
        threshold = 35  # 提高阈值，减少误检（原值20太敏感）
        candidates = np.where(diffs > threshold)[0]
        if len(candidates) > 0:
            return candidates[0] + 1
        return None

    num_samples = 5
    
    # Top Edge
    for i in range(1, num_samples + 1):
        x = int(roi_w * i / (num_samples + 1))
        col = roi_gray[:, x]
        w_val = scan_line(col, scan_limit)
        if w_val:
            detected_widths.append(w_val)
        
    # Bottom Edge
    for i in range(1, num_samples + 1):
        x = int(roi_w * i / (num_samples + 1))
        col = roi_gray[::-1, x]
        w_val = scan_line(col, scan_limit)
        if w_val:
            detected_widths.append(w_val)
        
    # Left Edge