{ "cells": [ { "metadata": { "ExecuteTime": { "end_time": "2025-04-14T04:06:31.135934Z", "start_time": "2025-04-14T04:06:30.845929Z" } }, "cell_type": "code", "source": [ "import cv2\n", "import numpy as np\n", "from skimage.feature import local_binary_pattern\n" ], "id": "9157e102c51206bb", "outputs": [], "execution_count": 1 }, { "metadata": { "ExecuteTime": { "end_time": "2025-04-14T04:06:31.144988Z", "start_time": "2025-04-14T04:06:31.138940Z" } }, "cell_type": "code", "source": [ "def extract_lbp_top(video_path, radius=2, n_points=8, method='uniform', block_size=10):\n", " cap = cv2.VideoCapture(video_path)\n", " frames = []\n", "\n", " # 读取所有帧\n", " while True:\n", " ret, frame = cap.read()\n", " if not ret:\n", " break\n", " gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)\n", " frames.append(gray)\n", " cap.release()\n", "\n", " frames = np.array(frames)\n", " T, H, W = frames.shape # 时间、空间维度\n", "\n", " # 使用滑动窗口计算 XT, YT 平面\n", " hist_xy = np.zeros((n_points + 2,))\n", " hist_xt = np.zeros((n_points + 2,))\n", " hist_yt = np.zeros((n_points + 2,))\n", "\n", " # LBP on XY plane\n", " for t in range(0, T, block_size):\n", " if t >= T:\n", " break\n", " lbp = local_binary_pattern(frames[t], n_points, radius, method)\n", " hist, _ = np.histogram(lbp.ravel(), bins=np.arange(0, n_points + 3), density=True)\n", " hist_xy += hist\n", "\n", " # LBP on XT plane\n", " for y in range(0, H, block_size):\n", " if y >= H:\n", " break\n", " xt_plane = frames[:, y, :]\n", " lbp = local_binary_pattern(xt_plane, n_points, radius, method)\n", " hist, _ = np.histogram(lbp.ravel(), bins=np.arange(0, n_points + 3), density=True)\n", " hist_xt += hist\n", "\n", " # LBP on YT plane\n", " for x in range(0, W, block_size):\n", " if x >= W:\n", " break\n", " yt_plane = frames[:, :, x]\n", " lbp = local_binary_pattern(yt_plane, n_points, radius, method)\n", " hist, _ = np.histogram(lbp.ravel(), bins=np.arange(0, n_points + 3), density=True)\n", " hist_yt += hist\n", "\n", " # 拼接三个平面的直方图作为最终特征向量\n", " feature_vector = np.concatenate([hist_xy, hist_xt, hist_yt])\n", " feature_vector /= np.linalg.norm(feature_vector) # 归一化\n", "\n", " return feature_vector" ], "id": "d9d5bd6f6a9e114a", "outputs": [], "execution_count": 2 }, { "metadata": { "ExecuteTime": { "end_time": "2025-04-14T04:06:31.297545Z", "start_time": "2025-04-14T04:06:31.294247Z" } }, "cell_type": "code", "source": [ "def getFeature(video_path = r\"D:\\DESKTOP\\2025\\44\\a1\\dataset\\org\\1.mp4\"):\n", "\n", " feature = extract_lbp_top(video_path)\n", " # print(\"Video feature shape (for copyright):\", feature.shape)\n", " return feature" ], "id": "404b8b3562026f1f", "outputs": [], "execution_count": 3 }, { "metadata": { "ExecuteTime": { "end_time": "2025-04-14T04:06:32.509232Z", "start_time": "2025-04-14T04:06:31.302931Z" } }, "cell_type": "code", "source": [ "import sqlite3\n", "import io\n", "import os\n", "import tqdm\n", "\n", "# 注册适配器与转换器：numpy数组 <-> BLOB\n", "def adapt_array(arr):\n", " out = io.BytesIO()\n", " np.save(out, arr)\n", " out.seek(0)\n", " return sqlite3.Binary(out.read())\n", "\n", "def convert_array(text):\n", " out = io.BytesIO(text)\n", " out.seek(0)\n", " return np.load(out)\n", "\n", "# 注册自定义类型处理\n", "sqlite3.register_adapter(np.ndarray, adapt_array)\n", "sqlite3.register_converter(\"array\", convert_array)\n", "\n", "# 创建数据库连接（带类型检测）\n", "conn = sqlite3.connect(\"LBPTOP.db\", detect_types=sqlite3.PARSE_DECLTYPES)\n", "cursor = conn.cursor()\n", "\n", "# 创建表\n", "cursor.execute(\"\"\"\n", "CREATE TABLE IF NOT EXISTS data (\n", " id INTEGER PRIMARY KEY,\n", " array BLOB,\n", " group_path TEXT,\n", " full_path TEXT\n", ")\n", "\"\"\")\n", "\n", "\n", "def add_to_db(array_to_store,group_path,full_path):\n", " cursor.execute(\"INSERT INTO data (array,group_path,full_path) VALUES (?,?,?)\", (array_to_store,group_path,full_path,))\n", " conn.commit()\n", "\n", "# # 读取数组\n", "# cursor.execute(\"SELECT array FROM data WHERE id=1\")\n", "# fetched_array = cursor.fetchone()[0]\n", "#\n", "# print(\"原始数组:\\n\", array_to_store)\n", "# print(\"读取的数组:\\n\", fetched_array)\n", "\n", "folder_path = r\"D:\\DESKTOP\\2025\\44\\a1\\dataset\"\n", "\n", "all_files = []\n", "names = [str(x)+\".mp4\" for x in range(10)]\n", "print(names)\n", "\n", "\n", "# 遍历文件夹\n", "for group_path in os.listdir(folder_path):\n", " full_path = os.path.join(folder_path, group_path)\n", " if os.path.isdir(full_path):\n", " # 遍历子文件夹\n", " for video_path in os.listdir(full_path):\n", " if os.path.basename(video_path) in names:\n", " video_full_path = os.path.join(full_path, video_path)\n", " if os.path.isfile(video_full_path):\n", " # 处理视频文件\n", " all_files.append((group_path,video_full_path))\n", "print(len(all_files))\n" ], "id": "4dc69bad309cb6b1", "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "['0.mp4', '1.mp4', '2.mp4', '3.mp4', '4.mp4', '5.mp4', '6.mp4', '7.mp4', '8.mp4', '9.mp4']\n", "70\n" ] } ], "execution_count": 4 }, { "metadata": { "ExecuteTime": { "end_time": "2025-04-14T04:20:52.302187Z", "start_time": "2025-04-14T04:06:32.521053Z" } }, "cell_type": "code", "source": [ "for group_path, video_full_path in tqdm.tqdm(all_files):\n", " # 读取视频特征\n", " feature = getFeature(video_full_path)\n", " # 将特征存储到数据库\n", " add_to_db(feature,group_path,video_full_path)\n", " # print(f\"已处理并存储: {video_full_path}\")\n", "\n", "conn.close()\n" ], "id": "3cd83672f6901125", "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "100%|██████████| 70/70 [14:19<00:00, 12.28s/it]\n" ] } ], "execution_count": 5 } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 2 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython2", "version": "2.7.6" } }, "nbformat": 4, "nbformat_minor": 5 }