我试图将高分辨率图像转换为更易于管理机器学习的东西。目前我有代码将图像大小调整为我想要的高度和宽度,但是我必须一次做一张图像,当我只做12-24张图像时我不得不放大,但很快我想扩大规模做几百张图片。我试图读取目录而不是单个图像,并将新图像保存在新目录中。初始图像将从.jpg,.png,.tif等变化,但我想将所有输出图像设置为.png,就像我在代码中一样。
import os
from PIL import Image
filename = "filename.jpg"
size = 250, 250
file_parts = os.path.splitext(filename)
outfile = file_parts[0] + '_250x250' + file_parts[1]
try:
img = Image.open(filename)
img = img.resize(size, Image.ANTIALIAS)
img.save(outfile, 'PNG')
except IOError as e:
print("An exception occured '%s'" %e)
任何有关此问题的帮助表示赞赏。
您可以使用循环遍历目录的内容
import os
for root, subdirs, files in os.walk(MY_DIRECTORY):
for f in files:
if f.endswith('png'):
#do something
假设您正在寻找的解决方案是同时处理多个图像 - 这是一个解决方案。有关更多信息,请参阅here。
from multiprocessing import Pool
def handle_image(image_file):
print(image_file)
#TODO implement the image manipulation here
if __name__ == '__main__':
p = Pool(5) # 5 as an example
# assuming you know how to prepare image file list
print(p.map(handle_image, ['a.jpg', 'b.jpg', 'c.png']))
你可以用这个:
#!/usr/bin/python
from PIL import Image
import os, sys
path = "\\path\\to\\files\\"
dirs = os.listdir( path )
def resize():
for item in dirs:
if os.path.isfile(path+item):
im = Image.open(path+item)
f, e = os.path.splitext(path+item)
imResize = im.resize((200,100), Image.ANTIALIAS)
imResize.save(f+'.png', 'png', quality=80)
resize()
您可以使用glob
浏览目录中的所有图像。然后使用opencv
调整图像大小如下,或者像PIL
一样。
import glob
import cv2
import numpy as np
IMG_DIR='home/xx/imgs'
def read_images(directory):
for img in glob.glob(directory+"/*.png"):
image = cv2.imread(img)
resized_img = cv2.resize(image/255.0 , (250 , 250))
yield resized_img
resized_imgs = np.array(list(read_images(IMG_DIR)))
我用了:
from PIL import Image
import os, sys
path = os.path.dirname(os.path.abspath(__file__))
dirs = os.listdir( path )
final_size = 244
print(dirs)
def resize_aspect_fit():
for item in dirs:
if ".PNG" in item:
print(item)
im = Image.open(path+"\\"+item)
f, e = os.path.splitext(path+"\\"+item)
size = im.size
print(size)
ratio = float(final_size) / max(size)
new_image_size = tuple([int(x*ratio) for x in size])
im = im.resize(new_image_size, Image.ANTIALIAS)
new_im = Image.new("RGB", (final_size, final_size))
new_im.paste(im, ((final_size-new_image_size[0])//2, (final_size-new_image_size[1])//2))
print(f)
new_im.save(f + 'resized.jpg', 'JPEG', quality=400)# png
resize_aspect_fit()