初始提交

onlyspectral_background.py

@@ -0,0 +1,162 @@
+import os
+import pandas as pd
+from pathlib import Path
+import numpy as np
+def process_single_bil_background(full_bil_path):
+    """
+    处理单个BIL文件，提取滤纸背景的平均光谱（不需要mask）
+
+    参数:
+        full_bil_path: BIL文件路径
+
+    返回:
+        mean_spectrum: 归一化的平均光谱数组
+    """
+    try:
+        from plantcv import plantcv as pcv
+        import cv2
+    except ImportError:
+        raise ImportError("需要导入plantcv和cv2库：from plantcv import plantcv as pcv; import cv2")
+
+    # 设置 PlantCV 的全局参数
+    pcv.params.debug = "None"
+    pcv.params.dpi = 100
+    pcv.params.text_size = 1
+    pcv.params.text_thickness = 1
+
+    # 读取光谱图像（ENVI 格式）
+    spectral_array = pcv.readimage(filename=full_bil_path, mode='envi')
+    bath_100 = spectral_array.array_data[:, :, 100]
+
+    # 使用阈值提取滤纸区域（背景区域，阈值>5500）
+    bath_over = pcv.threshold.binary(gray_img=bath_100, threshold=5500, object_type='light')
+
+    # 对滤纸掩膜进行形态学操作
+    bath_erode = pcv.erode(gray_img=bath_over, ksize=3, i=4)
+    dilate_img = pcv.dilate(gray_img=bath_erode, ksize=3, i=4)
+
+    # 填充大的孔洞
+    dilate_img = pcv.fill(bin_img=dilate_img, size=10000)
+
+    # 使用cv2.connectedComponents找到连通区域
+    num1, labeled_spectral_mask = cv2.connectedComponents(dilate_img.astype(np.uint8))
+
+    print(f"  检测到 {num1 - 1} 个滤纸区域")
+
+    # 找到最大面积的对象
+    max_area = 0
+    max_label = 0
+
+    for label in range(1, num1):
+        obj_mask = labeled_spectral_mask == label
+        area = np.sum(obj_mask)
+        if area > max_area:
+            max_area = area
+            max_label = label
+
+    # 提取最大面积对象的光谱
+    if max_label > 0:
+        max_obj_mask = labeled_spectral_mask == max_label
+        spectral_values = spectral_array.array_data[max_obj_mask, :]
+        mean_spectrum = np.mean(spectral_values, axis=0) / 10000  # 归一化
+        return mean_spectrum
+    else:
+        print(f"  警告: 未找到有效的滤纸区域")
+        n_bands = spectral_array.array_data.shape[-1]
+        return np.zeros(n_bands)
+
+
+def batch_process_bil_background(bil_folder, output_csv_path):
+    """
+    批量处理文件夹中的BIL文件，提取滤纸背景光谱并保存到CSV
+
+    参数:
+        bil_folder: 包含.bil文件的文件夹路径
+        output_csv_path: 输出CSV文件路径（宽格式：每行一个文件，每列一个波段）
+
+    返回:
+        df: 包含所有文件光谱的DataFrame
+    """
+    bil_folder = Path(bil_folder)
+    if not bil_folder.exists():
+        raise ValueError(f"文件夹不存在: {bil_folder}")
+
+    # 查找所有.bil文件
+    bil_files = []
+    for ext in ['.bil', '.BIL']:
+        bil_files.extend(list(bil_folder.glob(f'*{ext}')))
+
+    if len(bil_files) == 0:
+        print(f"警告: 在 {bil_folder} 中未找到任何.bil文件")
+        return None
+
+    print(f"找到 {len(bil_files)} 个BIL文件")
+
+    # 确保输出目录存在
+    output_path = Path(output_csv_path)
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+
+    # 存储所有结果
+    all_spectra = []
+    filenames = []
+    failed_files = []
+
+    # 处理每个文件
+    for i, bil_path in enumerate(bil_files, 1):
+        print(f"\n[{i}/{len(bil_files)}] 处理文件: {bil_path.name}")
+        print(f"{'=' * 60}")
+
+        try:
+            spectrum = process_single_bil_background(str(bil_path))
+
+            if spectrum is not None and len(spectrum) > 0 and np.sum(spectrum) > 0:
+                all_spectra.append(spectrum)
+                filenames.append(bil_path.stem)
+                print(f"  ✓ 成功提取光谱 ({len(spectrum)} 个波段)")
+            else:
+                failed_files.append(bil_path.name)
+                print(f"  ✗ 未能提取有效光谱")
+
+        except Exception as e:
+            failed_files.append(bil_path.name)
+            print(f"  ✗ 处理失败: {str(e)}")
+            import traceback
+            traceback.print_exc()
+            continue
+
+    # 合并结果
+    if len(all_spectra) == 0:
+        print("\n没有有效的结果可以保存")
+        return None
+
+    # 创建宽格式DataFrame
+    n_bands = len(all_spectra[0])
+    column_names = [f'band_{i}' for i in range(n_bands)]
+
+    df = pd.DataFrame(all_spectra, columns=column_names, index=filenames)
+    df.index.name = 'filename'
+    df = df.reset_index()
+
+    # 保存到CSV
+    df.to_csv(output_csv_path, index=False)
+
+    # 打印统计信息
+    print(f"\n{'=' * 60}")
+    print(f"批量处理完成！")
+    print(f"  总共文件数: {len(bil_files)}")
+    print(f"  成功处理: {len(all_spectra)}")
+    print(f"  失败文件: {len(failed_files)}")
+    if failed_files:
+        print(f"  失败列表: {', '.join(failed_files)}")
+    print(f"  光谱波段数: {n_bands}")
+    print(f"  输出文件: {output_csv_path}")
+    print(f"{'=' * 60}")
+
+    return df
+
+
+
+bil_folder = r"D:\Data\Traindata-09\whitebord"
+output_csv = r"D:\Data\Traindata-09\whitebord\filter_paper_spectra.csv"
+
+batch_process_bil_background(bil_folder, output_csv)