GreenTransPowerCalculate/deeplabv3sdRenewable/tools/山东省地貌识别tools/潜力评估阶段/面转栅格.py

import geopandas as gpd
import pandas as pd
import rasterio
from rasterio.features import rasterize
from rasterio.windows import Window
import numpy as np
from shapely.geometry import mapping
import os

def large_vector_to_raster(input_shp_path, output_raster_path, raster_resolution=30, chunk_size=1000, window_size=1000, category_field='value', category_value=1):
    """
    将大型面要素转为栅格数据，分块读取矢量并按窗口写入栅格，适用于大文件（例如2GB）

    参数:
    input_shp_path: 输入shapefile路径
    output_raster_path: 输出栅格文件路径
    raster_resolution: 输出栅格分辨率（单位：米，默认30）
    chunk_size: 每次读取的矢量特征数量（默认1000）
    window_size: 每个栅格窗口的像素大小（默认1000x1000）
    category_field: 矢量数据中用于栅格化的字段名（默认'value'）
    category_value: 默认类别值（若无字段，则所有特征赋此值，默认1）
    """
    try:
        # 检查输入文件是否存在
        if not os.path.exists(input_shp_path):
            raise FileNotFoundError(f"输入文件 {input_shp_path} 不存在")

        # 读取矢量文件的元信息以确定范围和CRS
        print("正在读取矢量元信息...")
        gdf_meta = gpd.read_file(input_shp_path, rows=1)  # 只读一行获取元数据
        base_crs = gdf_meta.crs
        bounds = gdf_meta.total_bounds  # [minx, miny, maxx, maxy]

        # 计算栅格尺寸
        width = int((bounds[2] - bounds[0]) / raster_resolution)
        height = int((bounds[3] - bounds[1]) / raster_resolution)
        transform = rasterio.transform.from_bounds(bounds[0], bounds[1], bounds[2], bounds[3], width, height)

        # 初始化输出栅格文件
        print(f"初始化输出栅格文件: {output_raster_path}")
        with rasterio.open(
            output_raster_path,
            'w',
            driver='GTiff',
            height=height,
            width=width,
            count=1,
            dtype=rasterio.uint8,
            crs=base_crs,
            transform=transform,
            nodata=0  # 背景值为0
        ) as dst:
            # 分窗口处理
            for row_offset in range(0, height, window_size):
                for col_offset in range(0, width, window_size):
                    window_height = min(window_size, height - row_offset)
                    window_width = min(window_size, width - col_offset)
                    window = Window(col_offset, row_offset, window_width, window_height)

                    # 计算当前窗口的地理范围
                    window_transform = rasterio.windows.transform(window, transform)
                    window_bounds = rasterio.windows.bounds(window, transform)

                    # 创建窗口的边界几何，用于筛选矢量数据
                    window_poly = gpd.GeoSeries.from_wkt([f"POLYGON(({window_bounds[0]} {window_bounds[1]}, "
                                                         f"{window_bounds[2]} {window_bounds[1]}, "
                                                         f"{window_bounds[2]} {window_bounds[3]}, "
                                                         f"{window_bounds[0]} {window_bounds[3]}, "
                                                         f"{window_bounds[0]} {window_bounds[1]}))"]).iloc[0]

                    # 分块读取矢量数据并筛选当前窗口内的特征
                    print(f"处理窗口: row={row_offset}, col={col_offset}")
                    chunk_raster = np.zeros((window_height, window_width), dtype=rasterio.uint8)
                    chunk_iterator = gpd.read_file(input_shp_path, chunksize=chunk_size)

                    for chunk_idx, chunk_gdf in enumerate(chunk_iterator):
                        # 筛选与当前窗口相交的特征
                        window_gdf = chunk_gdf[chunk_gdf.intersects(window_poly)]
                        if len(window_gdf) == 0:
                            continue

                        print(f"  - 处理分块 {chunk_idx}，筛选出 {len(window_gdf)} 个特征")
                        # 如果没有指定category_field，则赋默认值
                        if category_field not in window_gdf.columns:
                            window_gdf['value'] = category_value

                        # 栅格化当前分块
                        shapes = ((mapping(geom), value) for geom, value in zip(window_gdf.geometry, window_gdf[category_field]))
                        temp_raster = rasterize(
                            shapes=shapes,
                            out_shape=(window_height, window_width),
                            transform=window_transform,
                            fill=0,
                            dtype=rasterio.uint8
                        )

                        # 更新窗口栅格（保留非0值）
                        mask = temp_raster > 0
                        chunk_raster[mask] = temp_raster[mask]

                    # 写入当前窗口到栅格文件
                    if np.any(chunk_raster > 0):
                        dst.write(chunk_raster, 1, window=window)

        print("处理完成！")
        print(f"栅格保存为: {output_raster_path}")
        print(f"栅格中类别值: 0=背景, 其他值由 {category_field} 字段或默认值 {category_value} 确定")

    except Exception as e:
        print(f"发生错误: {str(e)}")


# 使用示例
if __name__ == "__main__":
    input_shp = r"E:\large_data\large_vector.shp"  # 替换为您的2GB矢量文件路径
    output_raster = r"E:\large_data\large_raster.tif"  # 输出栅格路径
    large_vector_to_raster(
        input_shp_path=input_shp,
        output_raster_path=output_raster,
        raster_resolution=30,  # 分辨率30米
        chunk_size=1000,       # 每次读取1000个特征
        window_size=1000,      # 每个窗口1000x1000像素
        category_field='value',  # 假设矢量数据中有'value'字段，若无则用默认值
        category_value=1       # 默认类别值
    )