Question

我能够从zip中成功加载图片：

with zipfile.ZipFile('test.zip', 'r') as zfile:
    data = zfile.read('test.jpg')
    # how to open this using imread or imdecode?

问题是：如何在不保存图像的情况下使用imread或imdecode在opencv中进行进一步处理？

更新

这是我得到的预期错误。我需要将'data'转换为opencv可以使用的类型。

data = zfile.read('test.jpg')
buf = StringIO.StringIO(data)
im = cv2.imdecode(buf, cv2.IMREAD_GRAYSCALE)
# results in error: TypeError: buf is not a numpy array, neither a scalar

a = np.asarray(buf)
cv2.imdecode(a, cv2.IMREAD_GRAYSCALE)
# results in error: TypeError: buf data type = 17 is not supported

Answer 1

使用numpy.frombuffer()从字符串创建uint8数组：

import zipfile
import cv2
import numpy as np

with zipfile.ZipFile('test.zip', 'r') as zfile:
    data = zfile.read('test.jpg')

img = cv2.imdecode(np.frombuffer(data, np.uint8), 1)

Answer 2

HYRY's answer确实提供了最优雅的解决方案

使用Python阅读图像并不完全符合"There should be one-- and preferably only one --obvious way to do it."

有时您可能宁愿避免在应用程序的某些部分使用numpy。并改用Pillow或imread。如果有一天你发现自己处于这种状况，那么希望下面的代码片段会有用：

import zipfile

with zipfile.ZipFile('test.zip', 'r') as zfile:
    data = zfile.read('test.jpg')

# Pillow
from PIL import Image
from StringIO import StringIO
import numpy as np
filelike_buffer = StringIO(data)
pil_image = Image.open(filelike_buffer)
np_im_array_from_pil = np.asarray(pil_image)
print type(np_im_array_from_pil), np_im_array_from_pil.shape
# <type 'numpy.ndarray'> (348, 500, 3)

# imread
from imread import imread_from_blob
np_im_array = imread_from_blob(data, "jpg")
print type(np_im_array), np_im_array.shape
# <type 'numpy.ndarray'> (348, 500, 3)

对"How to read raw png from an array in python opencv?"的回答提供了类似的解决方案。

使用python opencv从zip加载图像

2 个答案: