perf: 用 numpy 加速 is_blank_frame/compute_binary_matrix/compute_iou 三个像素遍历函数

doco/pyproject.toml

@@ -16,6 +16,7 @@ authors = [
 dependencies = [
     "Pillow>=10.0.0",
     "imagehash>=4.3.1",
+    "numpy>=1.24.0",
     "requests>=2.31.0",
     "python-dotenv>=1.0.0",
     "click>=8.1.0",

doco/src/doco/video_split.py

@@ -19,6 +19,7 @@ import tempfile
 from pathlib import Path
 from typing import Dict, List, Tuple, Optional
 
+import numpy as np
 from PIL import Image
 import imagehash
 
@@ -185,14 +186,12 @@ def is_blank_frame(image_path: Path, debug: bool = False) -> Tuple[bool, float]:
     返回: (is_blank, white_ratio)
     """
     img = Image.open(image_path).convert("L")
-    pixels = list(img.getdata())
+    arr = np.array(img)
-    total = len(pixels)
+    white_ratio = float(np.mean(arr > BLANK_FRAME_BRIGHTNESS_THRESHOLD))
-    white_count = sum(1 for p in pixels if p > BLANK_FRAME_BRIGHTNESS_THRESHOLD)
-    white_ratio = white_count / total if total > 0 else 0
     is_blank = white_ratio < BLANK_FRAME_WHITE_PIXEL_RATIO
 
     if debug:
-        frame_idx = image_path.stem  # e.g. "frame_0226"
+        frame_idx = image_path.stem
         print(f"[debug] {frame_idx}, white_ratio={white_ratio:.6f}, is_blank={is_blank}")
 
     return is_blank, white_ratio
@@ -226,39 +225,25 @@ def hamming_distance(s1: str, s2: str) -> int:
     return sum(c1 != c2 for c1, c2 in zip(s1, s2))
 
 
-def compute_binary_matrix(image_path: Path, threshold: int = 200) -> List[List[int]]:
+def compute_binary_matrix(image_path: Path, threshold: int = 200) -> np.ndarray:
     """
     将图片转为二值化矩阵(亮度 > threshold → 1,否则 → 0)
     用于 IoU 对比
+
+    返回: np.ndarray (dtype=uint8, 值为 0 或 1)
     """
     img = Image.open(image_path).convert("L")
-    w, h = img.size
+    arr = np.array(img)
-    matrix = []
+    return (arr > threshold).astype(np.uint8)
-    for y in range(h):
-        row = []
-        for x in range(w):
-            pixel = img.getpixel((x, y))
-            row.append(1 if pixel > threshold else 0)
-        matrix.append(row)
-    return matrix
 
 
-def compute_iou(matrix1: List[List[int]], matrix2: List[List[int]]) -> float:
+def compute_iou(matrix1: np.ndarray, matrix2: np.ndarray) -> float:
     """
     计算两个二值化矩阵的 IoU(交并比)
+    matrix1, matrix2: np.ndarray (dtype=uint8, 值为 0 或 1)
     """
-    h = len(matrix1)
+    intersection = int(np.sum((matrix1 == 1) & (matrix2 == 1)))
-    w = len(matrix1[0]) if h > 0 else 0
+    union = int(np.sum((matrix1 == 1) | (matrix2 == 1)))
-    intersection = 0
-    union = 0
-    for y in range(h):
-        for x in range(w):
-            v1 = matrix1[y][x]
-            v2 = matrix2[y][x]
-            if v1 == 1 or v2 == 1:
-                union += 1
-            if v1 == 1 and v2 == 1:
-                intersection += 1
     return intersection / union if union > 0 else 1.0