你就快成功了。
我假设您正在使用笔记本（否则，您需要知道如何呈现表格;在终端中，显然，这将是非常不同）
但是使用jupyter，您可以轻松地呈现HTML字符串，使用IPython
看到我自己的[mre]

import pandas as pd
from IPython.display import HTML
import numpy as np
from io import BytesIO
from PIL import Image
import base64

# A dataframe containing parameters for synthetic images
df=pd.DataFrame({'f1':[2,5], 'f2':[3, 1]})

# Some random images (I create the whole array of images)
# so a 4D array, with shape (len(df), 300, 300, 3)
# meaning len(df) 300×300 RGB images
x=np.linspace(0,2*np.pi,300)[None,None,:,None]
y=np.linspace(0,2*np.pi,300)[None,:,None,None]
phi=np.linspace(0,2*np.pi,3)[None,None,None,:]
f1=df.f1.values[:,None,None,None]
f2=df.f2.values[:,None,None,None]
imgs = np.sin(f1*x+phi)*np.sin(f2*y*phi) # Why not this

# So, now my example is set up. I have a dataframe with n rows. And n images.
# Let's fill the column with base 64 images.
# For this I need an image->string conversion
# that generates the image in HTML
# That is almost your code
def toB64(img):
    pimg=Image.fromarray((img*127+128).astype(np.uint8))
    my_stringIObytes = BytesIO()
    pimg.save(my_stringIObytes, format='png')
    my_stringIObytes.seek(0)
    b64 = base64.standard_b64encode(my_stringIObytes.read()).decode()
    return '<img src="data:image/png;base64,'+b64+'">' 
    # Difference is that I add the "data:image/png;base64,", to turn this base64 string into a "URL" (not a "link": the image is in the URL)
    # And that I add <img src=url> to turn this URL into an HTML image

# I can now create a new column containing the base64 images
df['imgs'] = [toB64(img) for img in imgs]

# And render the table
HTML(df.to_html(escape=False))

字符串
您缺少的两个重要行是toB64的最后一行，它从base64 URL创建img元素。最后一个是用HTML呈现表格（假设您在笔记本中）。escape=False意味着可能与html代码混淆的字符串内容不会像通常那样被转义，因此浏览器会将其视为真实的的HTML代码。
x1c 0d1x的数据