如何使用 Python 将图像转换为 NumPy 数组并保存为 CSV 文件?

原文:https://www . geesforgeks . org/如何使用 python 将图像转换为 numpy 数组并将其保存为 csv 文件/

让我们看看如何将图像转换成 NumPy 数组,然后用 Python 将该数组保存到 CSV 文件中?首先,我们将学习如何将图像转换为数字数组。将图像转换为数组的方法有很多种,其中很少几种是:

方法 1: 使用 PILNumPy 库。

我们将使用 PIL。Image.open() 和 numpy.asarray()

示例:

Python 3

# import required libraries
from PIL import Image
import numpy as gfg

# read an image
img = Image.open('geeksforgeeks.jpg')

# convert image object into array
imageToMatrice = gfg.asarray(img)

# printing shape of image
print(imageToMatrice.shape)

输出:

(251, 335, 3)

方法二:使用 Matplotlib 库。

我们将使用matplotlib . image . imread()方法。

示例:

Python 3

# import library
from matplotlib.image import imread

# read an image
imageToMatrice  = imread('geeksforgeeks.jpg')

# show shape of the image
print(imageToMatrice.shape)

输出:

(251, 335, 3)

现在,imageToMatrice 变量包含从给定图像转换后获得的数组。

获得的矩阵的维数取决于图像中存在多少通道:

  • 对于黑白或灰度图像:只存在一个通道,因此,矩阵的形状为(n,n),其中 n 代表图像的维度(像素),矩阵内的值范围为 0 到 255。
  • 对于颜色或 RGB 图像:它将渲染一个 3 通道的张量,因此矩阵的形状将是(n,n,3)。每个通道都是一个(n,n)矩阵,其中每个条目分别代表图像内实际位置的红色、绿色或蓝色级别。

我们将使用两种方法来做同样的事情,第一种方法使用 numpy 库,第二种方法使用 pandas 库:

注意:我们只能在文件中保存 1D 或 2D 矩阵,因此,灰度或黑白图像不会有问题,因为它是 2D 矩阵,但我们需要确保这适用于彩色或 RGB 图像,这是一个 3D 矩阵。

方法 1: 使用 NumPy 库。

我们将使用 numpy.savetxt()numpy.loadtxt()。

示例:

Python 3

# import required libraries
import numpy as gfg
import matplotlib.image as img

# read an image
imageMat = img.imread('gfg.jpg')
print("Image shape:", imageMat.shape)

# if image is colored (RGB)
if(imageMat.shape[2] == 3):

  # reshape it from 3D matrice to 2D matrice
  imageMat_reshape = imageMat.reshape(imageMat.shape[0],
                                      -1)
  print("Reshaping to 2D array:",
        imageMat_reshape.shape)

# if image is grayscale
else:
  # remain as it is
  imageMat_reshape = imageMat

# saving matrice to .csv file
gfg.savetxt('geek.csv',
            imageMat_reshape)

# retrieving matrice from the .csv file
loaded_2D_mat = gfg.loadtxt('geek.csv')

# reshaping it to 3D matrice
loaded_mat = loaded_2D_mat.reshape(loaded_2D_mat.shape[0],
                                   loaded_2D_mat.shape[1] // imageMat.shape[2],
                                   imageMat.shape[2])

print("Image shape of loaded Image:",
      loaded_mat.shape)

# check if both matrice have same shape or not
if((imageMat == loaded_mat).all()):

  print("\n\nYes",
        "The loaded matrice from CSV file is same as original image matrice")

输出:

图像形状:(251,335,3) 整形为 2D 阵列:(251,1005) 加载图像的图像形状:(251,335,3) 是 CSV 文件中加载的矩阵与原始图像矩阵相同

方法二:使用熊猫文库。

我们将使用熊猫。数据框()和熊猫。Dataframe()。to_csv() 法。

Python 3

# import required libraries
import numpy as gfg
import matplotlib.image as img
import pandas as pd

# read an image
imageMat = img.imread('gfg.jpg')
print("Image shape:",
      imageMat.shape)

# if image is colored (RGB)
if(imageMat.shape[2] == 3):

  # reshape it from 3D matrice to 2D matrice
  imageMat_reshape = imageMat.reshape(imageMat.shape[0],
                                      -1)
  print("Reshaping to 2D array:",
        imageMat_reshape.shape)

# if image is grayscale
else:
  # remain as it is
  imageMat_reshape = imageMat

# converting it to dataframe.
mat_df = pd.DataFrame(imageMat_reshape)

# exporting dataframe to CSV file.
mat_df.to_csv('gfgfile.csv',
              header = None,
              index = None)

# retrieving dataframe from CSV file
loaded_df = pd.read_csv('gfgfile.csv',
                        sep = ',',
                        header = None)
# getting matrice values.
loaded_2D_mat = loaded_df.values

# reshaping it to 3D matrice
loaded_mat = loaded_2D_mat.reshape(loaded_2D_mat.shape[0],
                                   loaded_2D_mat.shape[1] // imageMat.shape[2],
                                   imageMat.shape[2])

print("Image shape of loaded Image :",
      loaded_mat.shape)

# check if both matrice have same shape or not
if((imageMat == loaded_mat).all()):
  print("\n\nYes",
        "The loaded matrice from CSV file is same as original image matrice")

输出:

图像形状:(251,335,3) 整形为 2D 阵列:(251,1005) 加载图像的图像形状:(251,335,3) 是 CSV 文件中加载的矩阵与原始图像矩阵相同