批次對比結果集

Ding-yixia發表於2024-06-21
import os
import pickle

def compare_results_by_filename(file1, file2, threshold):
    # 載入兩個pickle檔案
    with open(file1, 'rb') as f:
        results1 = pickle.load(f)
    with open(file2, 'rb') as f:
        results2 = pickle.load(f)
    
    # 提取每個結果集中圖片的基本名字
    filenames1 = {os.path.splitext(os.path.basename(path))[0]: counts for path, counts in results1.items()}
    filenames2 = {os.path.splitext(os.path.basename(path))[0]: counts for path, counts in results2.items()}

    # 建立一個空列表來儲存差異大的圖片名稱
    significant_differences = []

    # 遍歷第一個結果集中的所有圖片名字
    for name, counts1 in filenames1.items():
        # 嘗試從第二個結果集中找到相同名字的圖片
        if name in filenames2:
            counts2 = filenames2[name]
            
            # 比較兩張圖片每個扇區的黑色畫素數量
            comparison = [abs(c1 - c2) for c1, c2 in zip(counts1, counts2)]
            
            # 檢查是否有扇區的差異超過閾值
            if any(d > threshold for d in comparison):
                significant_differences.append(name)
    
    return significant_differences

def batch_compare_directories(dir1, dir2, threshold):
    # 確保兩個目錄都存在
    assert os.path.exists(dir1), f"Directory {dir1} does not exist."
    assert os.path.exists(dir2), f"Directory {dir2} does not exist."

    # 獲取兩個目錄中的所有檔名
    files1 = set([f for f in os.listdir(dir1) if f.endswith('.pickle')])
    files2 = set([f for f in os.listdir(dir2) if f.endswith('.pickle')])

    # 確保兩個目錄中都有pickle檔案
    assert files1, f"No .pickle files found in {dir1}."
    assert files2, f"No .pickle files found in {dir2}."

    # 找出兩個目錄中都存在的pickle檔案
    common_files = files1.intersection(files2)

    # 建立一個空字典來儲存差異大的影像名稱
    all_diff_images = {}

    # 對於每一對pickle檔案，呼叫compare_results_by_filename函式
    for filename in common_files:
        file1 = os.path.join(dir1, filename)
        file2 = os.path.join(dir2, filename)
        
        # 呼叫函式比較兩個結果檔案
        diff_images = compare_results_by_filename(file1, file2, threshold)

        # 將差異大的影像名稱新增到字典中，鍵為pickle檔名
        if diff_images:
            all_diff_images[filename] = diff_images
    
    return all_diff_images

# 設定差異閾值
threshold = 100

# 呼叫函式比較兩個資料夾的結果
diff_images_dict = batch_compare_directories('FOLDER1', 'FOLDER2', threshold)

# 輸出有顯著差異的圖片名稱，按pickle檔案分組
if diff_images_dict:
    print("Images with significant differences in black pixel counts, grouped by pickle file:")
    for pickle_file, images in diff_images_dict.items():
        print(f"\nIn pickle file '{pickle_file}':")
        for img in images:
            print(img)
else:
    print("No significant differences found.")

相關文章