python中RAW图像的JPEG解压

Question

当我尝试对 RAW 数据进行 JPEG 解压缩（旧式 JPEG 压缩，而不是 JPEG-LS 和 JPEG2000）时，出现以下错误：

Traceback (most recent call last):
  File "raw-reader.py", line 766, in <module>
    raw_image_data = imageio.imread(io.BytesIO(raw_packed_image_data))
  File "/home/ian/.local/lib/python3.6/site-packages/imageio/core/functions.py", line 206, in imread
    reader = read(uri, format, 'i', **kwargs)
  File "/home/ian/.local/lib/python3.6/site-packages/imageio/core/functions.py", line 129, in get_reader
    return format.get_reader(request)
  File "/home/ian/.local/lib/python3.6/site-packages/imageio/core/format.py", line 168, in get_reader
    return self.Reader(self, request)
  File "/home/ian/.local/lib/python3.6/site-packages/imageio/core/format.py", line 217, in __init__
    self._open(**self.request.kwargs.copy())
  File "/home/ian/.local/lib/python3.6/site-packages/imageio/plugins/pillow.py", line 398, in _open
    pilmode=pilmode, as_gray=as_gray)
  File "/home/ian/.local/lib/python3.6/site-packages/imageio/plugins/pillow.py", line 122, in _open
    self._im = factory(self._fp, '')
  File "/home/ian/.local/lib/python3.6/site-packages/PIL/JpegImagePlugin.py", line 780, in jpeg_factory
    im = JpegImageFile(fp, filename)
  File "/home/ian/.local/lib/python3.6/site-packages/PIL/ImageFile.py", line 102, in __init__
    self._open()
  File "/home/ian/.local/lib/python3.6/site-packages/PIL/JpegImagePlugin.py", line 339, in _open
    handler(self, i)
  File "/home/ian/.local/lib/python3.6/site-packages/PIL/JpegImagePlugin.py", line 166, in SOF
    raise SyntaxError("cannot handle %d-bit layers" % self.bits)
SyntaxError: cannot handle 14-bit layers

图像中的 RAW 数据是 14 位 JPEG 数据，imageio 无法读取它。当我尝试使用枕头时，它甚至无法将数据识别为 JPEG。我现在的问题是：如何在不编写自己的 JPEG 解压缩器的情况下解压缩数据，同时记住数据是 14 位的？

我的代码：

import io
import imageio

allbytes = open("raw_data.dat", "rb").read()

raw_packed_image_data = allbytes
raw_image_data = imageio.imread(io.BytesIO(raw_packed_image_data))

文件raw_data.dat 是一个纯粹包含用JPEG 压缩的RAW-Image 数据的文件。链接到raw_data.dat

Answer 1

raw_data.dat 是一个 JPEG Lossless, Nonhierarchical 文件，有 2 帧，精度 > 8 位，非常罕见的格式。

imagecodecs 包可以读取文件（假设存在_imagecodecs Cython 扩展）：

>>> from imagecodecs import jpegsof3_decode
>>> data = open('raw_data.dat', 'rb').read()
>>> image = jpegsof3_decode(data)
>>> image.shape
(3528, 2640, 2)
>>> image.dtype
dtype('uint16')

LEADTOOLS SDK 也应该能够读取文件（未测试）。

Answer 2

我相信可以使用不同的库来加载图像来解决这个问题。类似的潜在问题已在此处发布How to combine 3 high range JPEG2000 images into single RGB one?。您可以使用以下方法：

import matplotlib.image as mpimg
img_red = mpimg.imread('raw_data.dat')

然后您可以使用读取的位图进行进一步操作，就好像它是通过 PIL/Pillow/imageio 加载的一样。