229 lines
11 KiB
Python
229 lines
11 KiB
Python
import geopandas as gpd
|
||
import pandas as pd
|
||
import rasterio
|
||
from rasterio.features import rasterize
|
||
import numpy as np
|
||
from shapely.geometry import mapping, box
|
||
import os
|
||
from tqdm import tqdm # 导入 tqdm 用于进度条
|
||
|
||
|
||
def merge_vectors_seven_categories(file_paths, output_vector_path):
|
||
"""
|
||
合并7个地貌类型矢量图层并保存为GeoPackage格式（去掉cropland）
|
||
|
||
参数:
|
||
file_paths: 包含7个矢量文件路径的列表 (按顺序: forest, water, shrub, wetland, jianzhu, grass, bareland)
|
||
output_vector_path: 输出合并后的矢量文件路径（使用.gpkg格式）
|
||
"""
|
||
try:
|
||
# 检查输入文件数量是否为7
|
||
if len(file_paths) != 7:
|
||
raise ValueError("请提供正好7个矢量文件路径")
|
||
|
||
for path in file_paths:
|
||
if not os.path.exists(path):
|
||
raise FileNotFoundError(f"文件不存在: {path}")
|
||
|
||
# 定义7个类别（去掉cropland）
|
||
categories = ['forest', 'water', 'shrub', 'wetland', 'jianzhu', 'grass', 'bareland']
|
||
category_values = {cat: i + 1 for i, cat in enumerate(categories)} # forest=1, water=2, ..., bareland=7
|
||
|
||
gdfs = []
|
||
for i, (path, category) in enumerate(zip(file_paths, categories), 1):
|
||
print(f"正在读取第{i}个矢量文件: {path}")
|
||
gdf = gpd.read_file(path, engine="pyogrio")
|
||
gdf['category'] = category
|
||
gdf['value'] = category_values[category]
|
||
if 'Shape_Area' in gdf.columns:
|
||
gdf['Shape_Area'] = gdf['Shape_Area'].clip(upper=1e9)
|
||
gdfs.append(gdf)
|
||
|
||
# 统一坐标系
|
||
base_crs = gdfs[0].crs
|
||
for i, gdf in enumerate(gdfs[1:], 2):
|
||
if gdf.crs != base_crs:
|
||
print(f"第{i}个图层坐标系不同，正在转换为第一个图层的坐标系...")
|
||
gdfs[i - 1] = gdf.to_crs(base_crs)
|
||
|
||
print("正在合并图层...")
|
||
merged_gdf = gpd.GeoDataFrame(pd.concat(gdfs, ignore_index=True))
|
||
merged_gdf = merged_gdf.set_geometry('geometry')
|
||
|
||
if not output_vector_path.endswith('.gpkg'):
|
||
output_vector_path = output_vector_path.replace('.shp', '.gpkg')
|
||
print("输出格式已更改为GeoPackage以支持大文件")
|
||
print(f"正在保存合并矢量结果到: {output_vector_path}")
|
||
merged_gdf.to_file(output_vector_path, driver='GPKG', engine="pyogrio")
|
||
|
||
print("矢量合并完成！")
|
||
print(f"合并矢量保存为: {output_vector_path}")
|
||
return output_vector_path
|
||
|
||
except Exception as e:
|
||
print(f"发生错误: {str(e)}")
|
||
return None
|
||
|
||
|
||
def rasterize_vector_by_blocks(input_vector_path, output_raster_path, raster_resolution=30, block_size=5000):
|
||
"""
|
||
将合并后的矢量文件分块转为栅格，显示7个类别，背景值设为 value=8，添加进度条和像素统计
|
||
|
||
参数:
|
||
input_vector_path: 输入的合并矢量文件路径（.gpkg格式）
|
||
output_raster_path: 输出栅格文件路径
|
||
raster_resolution: 输出栅格的分辨率（默认30米）
|
||
block_size: 每个分块的宽度和高度（像素单位，默认5000）
|
||
"""
|
||
try:
|
||
# 读取合并后的矢量文件
|
||
print(f"正在读取合并矢量文件: {input_vector_path}")
|
||
merged_gdf = gpd.read_file(input_vector_path, engine="pyogrio")
|
||
|
||
# 检查 value 字段是否有效
|
||
if 'value' not in merged_gdf.columns:
|
||
raise ValueError("矢量文件中缺少 'value' 字段，请确保文件包含正确的分类值")
|
||
print("合并矢量文件中的 value 字段分布:")
|
||
print(merged_gdf['value'].value_counts(dropna=False))
|
||
|
||
if not merged_gdf['value'].between(1, 7).all():
|
||
print("警告: 'value' 字段包含无效值（应在 1-7 之间），可能导致数据丢失")
|
||
merged_gdf = merged_gdf[merged_gdf['value'].between(1, 7)]
|
||
|
||
# 检查合并后的矢量数据是否为空
|
||
if merged_gdf.empty:
|
||
raise ValueError("合并后的矢量数据为空，请检查输入数据是否有效")
|
||
|
||
# 按 value 字段排序，优先级高的类别（例如 value 较大）后处理，避免被覆盖
|
||
print("正在按 'value' 字段排序以确保优先级...")
|
||
merged_gdf = merged_gdf.sort_values(by='value', ascending=True)
|
||
|
||
# 计算总范围和栅格尺寸
|
||
bounds = merged_gdf.total_bounds # [minx, miny, maxx, maxy]
|
||
total_width = int((bounds[2] - bounds[0]) / raster_resolution)
|
||
total_height = int((bounds[3] - bounds[1]) / raster_resolution)
|
||
|
||
if total_width <= 0 or total_height <= 0:
|
||
raise ValueError(f"栅格尺寸无效: 宽度={total_width}, 高度={total_height}")
|
||
|
||
print(f"栅格范围: minx={bounds[0]}, miny={bounds[1]}, maxx={bounds[2]}, maxy={bounds[3]}")
|
||
print(f"栅格尺寸: 宽度={total_width}, 高度={total_height}")
|
||
|
||
# 创建栅格变换
|
||
transform = rasterio.transform.from_bounds(bounds[0], bounds[1], bounds[2], bounds[3], total_width,
|
||
total_height)
|
||
|
||
# 计算总分块数以设置进度条
|
||
total_blocks = ((total_height + block_size - 1) // block_size) * ((total_width + block_size - 1) // block_size)
|
||
print(f"总分块数: {total_blocks}")
|
||
|
||
# 创建并写入栅格文件，逐块写入以减少内存占用
|
||
with rasterio.open(
|
||
output_raster_path,
|
||
'w',
|
||
driver='GTiff',
|
||
height=total_height,
|
||
width=total_width,
|
||
count=1,
|
||
dtype=rasterio.uint8,
|
||
crs=merged_gdf.crs,
|
||
transform=transform,
|
||
nodata=255 # 使用 255 作为 nodata 值，区分 value=8 的背景
|
||
) as dst:
|
||
# 分块处理并添加进度条
|
||
with tqdm(total=total_blocks, desc="栅格化进度") as pbar:
|
||
for y in range(0, total_height, block_size):
|
||
for x in range(0, total_width, block_size):
|
||
block_height = min(block_size, total_height - y)
|
||
block_width = min(block_size, total_width - x)
|
||
|
||
# 计算当前分块的地理范围
|
||
block_minx = bounds[0] + x * raster_resolution
|
||
block_maxy = bounds[3] - y * raster_resolution
|
||
block_maxx = block_minx + block_width * raster_resolution
|
||
block_miny = block_maxy - block_height * raster_resolution
|
||
|
||
# 创建分块边界框
|
||
block_bounds = box(block_minx, block_miny, block_maxx, block_maxy)
|
||
block_gdf = merged_gdf[merged_gdf.geometry.intersects(block_bounds)].copy()
|
||
|
||
# 初始化分块为背景值 8
|
||
block_raster = np.full((block_height, block_width), 8, dtype=np.uint8)
|
||
|
||
if not block_gdf.empty:
|
||
# 打印分块中的 value 字段分布
|
||
print(f"分块 (x={x}, y={y}) 中的 value 字段分布:")
|
||
print(block_gdf['value'].value_counts(dropna=False))
|
||
|
||
# 分块栅格化
|
||
block_transform = rasterio.transform.from_bounds(
|
||
block_minx, block_miny, block_maxx, block_maxy, block_width, block_height
|
||
)
|
||
shapes = ((mapping(geom), value) for geom, value in
|
||
zip(block_gdf.geometry, block_gdf['value']))
|
||
block_raster = rasterize(
|
||
shapes=shapes,
|
||
out=block_raster, # 使用预初始化的数组
|
||
transform=block_transform,
|
||
fill=8, # 未覆盖区域为 8
|
||
dtype=rasterio.uint8,
|
||
all_touched=True # 确保所有触及的像素都被计入
|
||
)
|
||
|
||
# 统计分块中 value=1 到 value=7 的像素点个数
|
||
print(f"分块 (x={x}, y={y}) 中像素值统计:")
|
||
unique, counts = np.unique(block_raster, return_counts=True)
|
||
value_counts = dict(zip(unique, counts))
|
||
for val in range(1, 8): # 检查 value=1 到 value=7
|
||
count = value_counts.get(val, 0)
|
||
print(f"Value={val}: {count} 像素")
|
||
# 单独打印 value=8（背景值）
|
||
background_count = value_counts.get(8, 0)
|
||
print(f"Value=8 (background): {background_count} 像素")
|
||
|
||
# 直接写入当前分块到文件中
|
||
dst.write(block_raster, 1, window=((y, y + block_height), (x, x + block_width)))
|
||
print(f"已完成分块栅格化: x={x}, y={y}, 宽度={block_width}, 高度={block_height}")
|
||
pbar.update(1)
|
||
|
||
# 读取整个栅格文件，统计最终像素分布
|
||
with rasterio.open(output_raster_path) as src:
|
||
final_raster = src.read(1)
|
||
print("最终栅格文件中的像素值统计:")
|
||
unique, counts = np.unique(final_raster, return_counts=True)
|
||
value_counts = dict(zip(unique, counts))
|
||
for val in range(1, 9): # 检查 value=1 到 value=8
|
||
count = value_counts.get(val, 0)
|
||
print(f"Value={val}: {count} 像素")
|
||
# 打印 nodata 值（255）
|
||
nodata_count = value_counts.get(255, 0)
|
||
print(f"Nodata (255): {nodata_count} 像素")
|
||
|
||
print("栅格化完成！")
|
||
print(f"栅格保存为: {output_raster_path}")
|
||
print(f"栅格中类别值: 1=forest, 2=water, 3=shrub, 4=wetland, 5=jianzhu, 6=grass, 7=bareland, 8=background")
|
||
|
||
except Exception as e:
|
||
print(f"发生错误: {str(e)}")
|
||
|
||
|
||
# 使用示例
|
||
if __name__ == "__main__":
|
||
input_files = [
|
||
r"sddimaoshp/forest/sdforest.shp",
|
||
r"sddimaoshp/water/shandongsuiyu3.shp",
|
||
r"sddimaoshp/shrub/sdshrub.shp",
|
||
r"sddimaoshp/wetland/sdwetland.shp",
|
||
r"sddimaoshp/jianzhu/shandongjianzhu3.shp",
|
||
r"sddimaoshp/grass/sdgrass.shp",
|
||
r"sddimaoshp/bareland/sdbareland.shp"
|
||
]
|
||
output_vector_file = r"sddimaoshp/hebingtif/shandongdimiao_10_1m.gpkg"
|
||
output_raster_file = r"sddimaoshp/hebingtif/sddimao_10_1m.tif"
|
||
|
||
# Step 1: 合并矢量
|
||
vector_path = merge_vectors_seven_categories(input_files, output_vector_file)
|
||
|
||
# Step 2: 如果合并成功，进行分块栅格化
|
||
if vector_path:
|
||
rasterize_vector_by_blocks(vector_path, output_raster_file, raster_resolution=10, block_size=5000) |