从numpy数组创建光栅，从csv fi获取值

Question

我有个朋友。我想用csv表中相应的值替换光栅中的值。你知道吗

光栅的类值为0到n，csv为光栅的每个类n都有一个计算值（例如点密度）。 我想从csv中的相应值创建一个新光栅

我正在使用GDAL和numpy。我试过使用pandas，但遇到了从csv提取值到光栅pandas数据帧的问题。我将在对应csv表的光栅列表上执行此操作。你知道吗

下面是我的数据示例（一个光栅）

#Example raster array
[5 2 2 3
 0 3 1 4
 2 0 1 3]

#Corresponding csv table
  Class   Count  Density
    0       2       6
    1       2       9
    2       2       4
    3       3       9
    4       1       7
    5       1       2


#Output Raster (to take the corresponding density values, 
#i.e. if class = 0, then output raster = 6, the corresponding density value)
    [2 4 4 9
     6 9 9 7
     4 6 9 9]

我有从光栅创建数组和从数组写回光栅的代码。我从不同的stackexchange网站上发现的。 我不知道如何框循环，以获得新光栅中的csv值。 我下面的for循环代码不完整。 有人能帮忙吗

import numpy, sys
from osgeo import gdal
from osgeo.gdalconst import *

inRst = gdal.Open(r"c:/Raster1.tif")
band = inRst.GetRasterBand(1)
rows = inRst.RasterYSize
cols = inRst.RasterXSize
rstr_arry = band.ReadAsArray(0,0,cols,rows)

# create the output image
driver = inRst.GetDriver()
#print driver
outRst = driver.Create(r"c:/NewRstr.tif", cols, rows, 1, GDT_Int32)
outBand = outRst.GetRasterBand(1)
outData = numpy.zeros((rows,cols), numpy.int32)

for i in range(0, rows):
    for j in range(0, cols):
        if rstr_arry[i,j] =  :
            outData[i,j] = 
        elif rstr_arry[i,j] = :
            outData[i,j] = 
        else:
            outData[i,j] = 


# write the data
outRst= outBand.WriteArray(outData, 0, 0)
# flush data to disk, set the NoData value and calculate stats
outBand.FlushCache()
outBand.SetNoDataValue(-99)
# georeference the image and set the projection
outDs.SetGeoTransform(inDs.GetGeoTransform())
outDs.SetProjection(inDs.GetProjection())

Answer 1

如果我没有误解您想要实现的目标，那么您首先必须读取csv文件并创建Class值到Density值的映射。可以这样做：

import csv

mapping = {}

with open('test.csv') as csv_file:
    csv_reader = csv.DictReader(csv_file)
    for row in csv_reader:
        mapping[int(row['Class'])] = int(row['Density'])

您将获得如下dict：

{0: 6, 1: 9, 2: 4, 3: 9, 4: 7, 5: 2}

然后可以使用^{}创建需要替换的元素的掩码矩阵，并使用^{}替换元素。在执行此操作之前，需要展平光栅阵列，并在写回结果之前恢复其形状。 （替换numpy数组中元素的替代方法可以在以下问题的答案中找到：Fast replacement of values in a numpy array）

# Save the shape of the raster array
s = rstr_arry.shape
# Flatten the raster array
rstr_arry = rstr_arry.reshape(-1)
# Create 2D replacement matrix:
replace = numpy.array([list(mapping.keys()), list(mapping.values())])
# Find elements that need replacement:
mask = numpy.in1d(rstr_arry, replace[0, :])
# Replace them:
rstr_arry[mask] = replace[1, numpy.searchsorted(replace[0, :], rstr_arry[mask])]
# Restore the shape of the raster array
rstr_arry = rstr_arry.reshape(s)

然后，您可以按照计划编写数据几乎：

outBand.WriteArray(rstr_arry, 0, 0)
outBand.SetNoDataValue(-99)

outDs.SetGeoTransform(inRst.GetGeoTransform())
outDs.SetProjection(inRst.GetProjection())

outBand.FlushCache()

---

在示例数据上进行测试：

rstr_arry = np.asarray([
    [5, 2, 2, 3],
    [0, 3, 1, 4],
    [2, 0, 1, 3]])

mapping = {0: 6, 1: 9, 2: 4, 3: 9, 4: 7, 5: 2}

s = rstr_arry.shape
rstr_arry = rstr_arry.reshape(-1)
replace = numpy.array([list(mapping.keys()), list(mapping.values())])
mask = numpy.in1d(rstr_arry, replace[0, :])
rstr_arry[mask] = replace[1, numpy.searchsorted(replace[0, :], rstr_arry[mask])]
rstr_arry = rstr_arry.reshape(s)

print(rstr_arry)
# [[2 4 4 9]
#  [6 9 9 7]
#  [4 6 9 9]]