请你帮我写出修改后的完整代码

当然，这里是经过修改后的完整代码，包含了`process_grid_for_resnet`函数用于处理单个网格的特征提取以及新的`extract_features`函数处理整个栅格列表：

import numpy as np
import torch
from torchvision import models, transforms
from PIL import Image
import pandas as pd

# 其他代码不变

# 图像预处理
preprocess = transforms.Compose([
    transforms.Resize((224, 224)),  # 使用固定大小以保持ResNet的一致性
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])

# 定义处理单个网格的函数
def process_grid_for_resnet(grid_tensor, resnet, preprocess):
    with torch.no_grad():
        grid_tensor = preprocess(grid_tensor)
        grid_tensor = grid_tensor.unsqueeze(0)  # 添加 batch dimension
        feature_map = resnet(grid_tensor)  # 前向传播
        # 获取最后一层的特征图，通常是卷积层的结果
        features = feature_map[:, :, :, :].squeeze(0)  # 按需选择特定层的特征
        return features

# 提取特征函数
def extract_features(grids, resnet, preprocess):
    features_list = []
    for grid in grids:
        features = process_grid_for_resnet(grid, resnet, preprocess)
        features_list.append(features)
    return features_list

# 加载并划分图像
poi_grids = load_and_split_image(poi_image_path, grid_width, grid_height)
nightlight_grids = load_and_split_image(nightlight_image_path, grid_width, grid_height)

# 提取特征
poi_features = extract_features(poi_grids, resnet, preprocess)
nightlight_features = extract_features(nightlight_grids, resnet, preprocess)

# 融合特征
def fuse_features(poi_features, nightlight_features, method='sum'):
    fused_features = []
    for poi_feat, nightlight_feat in zip(poi_features, nightlight_features):
        if method == 'sum':
            fused_feature = poi_feat + nightlight_feat
        elif method == 'average':
            fused_feature = (poi_feat + nightlight_feat) / 2
        else:
            raise ValueError("Invalid fusion method. Choose 'sum' or 'average'.")
        fused_features.append(fused_feature)
    return fused_features

fused_features = fuse_features(poi_features, nightlight_features, 'average')  # 或者改为 'sum'

# 输出融合后的特征值列表
fused_feature_values = [feature.tolist() for feature in fused_features]

# 转换为 DataFrame
df = pd.DataFrame(fused_feature_values, columns=['Fused Feature'])

# 保存到 Excel 文件
df.to_excel('fused_features.xlsx', index=False)

在这个版本中，我添加了一个`fuse_features`函数，它接受一个融合方法（默认为求和，也可以改选为平均）。根据您选择的融合方法，您可以调用该函数对 `poi_features` 和 `nightlight_features` 进行融合。