PIL 不读取DNG文件,也不喜欢每通道16位的RGB。我想你需要这样的东西 rawpy

import rawpy

path = '/Users/mark/Desktop/RAW_LEICA_M8.DNG'
with rawpy.imread(path) as raw:
    rgb = raw.postprocess()

print(rgb.shape)       # It's a Numpy array

三倍输出

Out[9]: (2638, 3920, 3)

另一个选择是 Exiftool ,您可以这样安装:

pip3 install PyExifTool

然后像这样使用:

import exiftool

with exiftool.ExifTool() as et:
    metadata = et.get_metadata('/path/to/RAW_LEICA_M8.DNG')

print(metadata)

{'SourceFile': '/path/to/RAW_LEICA_M8.DNG',
 'ExifTool:ExifToolVersion': 12.0,
 'File:FileName': 'RAW_LEICA_M8.DNG',
 'File:Directory': '/path/to',
 'File:FileSize': 10575296,
 'File:FileModifyDate': '2021:08:16 08:54:30+01:00',
 'File:FileAccessDate': '2021:08:16 09:56:33+01:00',
 'File:FileInodeChangeDate': '2021:08:16 09:56:31+01:00',
 'File:FilePermissions': 644,
 'File:FileType': 'DNG',
 'File:FileTypeExtension': 'DNG',
 'File:MIMEType': 'image/x-adobe-dng',
 'File:ExifByteOrder': 'II',
 'EXIF:Make': 'Leica Camera AG',
 'EXIF:Model': 'M8 Digital Camera',
 'EXIF:Orientation': 1,
 'EXIF:Software': 1.107,
 'EXIF:Artist': '',
 'EXIF:SubfileType': 0,
 'EXIF:ImageWidth': 3920,            <--- HERE IS THE WIDTH
 'EXIF:ImageHeight': 2638,           <--- HERE IS THE HEIGHT
 'EXIF:BitsPerSample': 8,
 'EXIF:Compression': 1,
 'EXIF:PhotometricInterpretation': 32803,
 'EXIF:StripOffsets': '(Binary data 1315 bytes, use -b option to extract)',
 'EXIF:SamplesPerPixel': 1,
 'EXIF:RowsPerStrip': 16,
 'EXIF:StripByteCounts': '(Binary data 989 bytes, use -b option to extract)',
 'EXIF:XResolution': 300,
 'EXIF:YResolution': 300,
 'EXIF:PlanarConfiguration': 1,
 'EXIF:ResolutionUnit': 2,
 'EXIF:CFARepeatPatternDim': '2 2',
 'EXIF:CFAPattern2': '0 1 1 2',
 'EXIF:LinearizationTable': '(Binary data 1244 bytes, use -b option to extract)',
 'EXIF:WhiteLevel': 16383,
 'EXIF:DefaultCropOrigin': '2 2',
 'EXIF:DefaultCropSize': '3916 2634',
 'EXIF:BayerGreenSplit': 500,
 'EXIF:AntiAliasStrength': 0,
 'EXIF:Copyright': '',
 'EXIF:ExposureTime': 12,
 'EXIF:ExposureProgram': 1,
 'EXIF:ISO': 160,
 'EXIF:ExifVersion': '0220',
 'EXIF:CreateDate': '2007:08:02 22:13:49',
 'EXIF:ShutterSpeedValue': 11.3137084989848,
 'EXIF:ExposureCompensation': 0,
 'EXIF:MaxApertureValue': 2,
 'EXIF:MeteringMode': 2,
 'EXIF:LightSource': 0,
 'EXIF:Flash': 0,
 'EXIF:FocalLength': 50,
 'EXIF:FileSource': 3,
 'EXIF:SceneType': 1,
 'EXIF:WhiteBalance': 0,
 'EXIF:DigitalZoomRatio': 0,
 'EXIF:FocalLengthIn35mmFormat': 67,
 'EXIF:SceneCaptureType': 0,
 'EXIF:ImageUniqueID': '00000000000000000000000000000147',
 'EXIF:SelfTimerMode': 0,
 'EXIF:DateTimeOriginal': '2007:08:02 22:13:49',
 'EXIF:FocalPlaneXResolution': 3729,
 'EXIF:FocalPlaneYResolution': 3764,
 'EXIF:FocalPlaneResolutionUnit': 2,
 'EXIF:TIFF-EPStandardID': '0 0 0 1',
 'EXIF:DNGVersion': '1 0 0 0',
 'EXIF:UniqueCameraModel': 'M8 Digital Camera',
 'EXIF:ColorMatrix1': '1.0469 -0.5314 0.128 -0.4326 1.2176 0.2419 -0.0886 0.2473 0.716',
 'EXIF:ColorMatrix2': '0.7675 -0.2195 -0.0305 -0.586 1.4118 0.1857 -0.2425 0.4007 0.6578',
 'EXIF:CameraCalibration1': '1 0 0 0 1 0 0 0 1',
 'EXIF:CameraCalibration2': '1 0 0 0 1 0 0 0 1',
 'EXIF:AsShotNeutral': '0.4750637903 1 0.7966159382',
 'EXIF:BaselineNoise': 1,
 'EXIF:BaselineSharpness': 1,
 'EXIF:CameraSerialNumber': 3106091,
 'EXIF:CalibrationIlluminant1': 17,
 'EXIF:CalibrationIlluminant2': 21,
 'MakerNotes:UserProfile': 1,
 'MakerNotes:SerialNumber': 3106091,
 'MakerNotes:WhiteBalance': 0,
 'MakerNotes:LensType': '33 3',
 'MakerNotes:ExternalSensorBrightnessValue': -5.16796875,
 'MakerNotes:MeasuredLV': -4.359375,
 'MakerNotes:ApproximateFNumber': 4,
 'MakerNotes:CameraTemperature': 19,
 'MakerNotes:ColorTemperature': 5515,
 'MakerNotes:UV-IRFilterCorrection': 0,
 'MakerNotes:CCDVersion': 0,
 'MakerNotes:CCDBoardVersion': 1,
 'MakerNotes:ControllerBoardVersion': 0,
 'MakerNotes:M16CVersion': 0,
 'MakerNotes:ImageIDNumber': 327,
 'Composite:CFAPattern': '2 2 0 1 1 2',
 'Composite:ImageSize': '3920 2638',
 'Composite:LensID': '33 3',
 'Composite:Megapixels': 10.34096,
 'Composite:ScaleFactor35efl': 1.34,
 'Composite:ShutterSpeed': 12,
 'Composite:CircleOfConfusion': 0.0224225825588557,
 'Composite:FOV': 30.0756572109854,
 'Composite:FocalLength35efl': 67}

从下载的原始图像示例 Raw Samples .

如果您的图像是1024x 1024像素,那么如果数据是8位的,那么这将生成1048576字节。但是你的文件是2097268字节,这仅仅是预期大小的两倍多一点,所以我猜你的数据是16位的,即每像素2字节。这意味着文件中有2097268-(2*1024*1024),即116字节的其他垃圾。人们通常在文件的开头存储额外的内容。所以,我只取了文件的最后2097152字节,并假设它是一个16位的灰度图像,大小为1024x1024。

您可以使用ImageMagick在终端的命令行中执行此操作,如下所示:

magick -depth 16 -size 1024x1024+116 gray:frame_064_0000.img -auto-level result.png

在Python中,可以打开文件,从文件末尾向后查找2097152字节,然后将其读入uint16的1024x102np.array。

看起来像这样:

import numpy as np
from PIL import Image

filename = 'frame_064_0000.img' 

# set width and height 
w, h = 1024, 1024 

with open(filename, 'rb') as f: 
    # Seek backwards from end of file by 2 bytes per pixel 
    f.seek(-w*h*2, 2) 
    img = np.fromfile(f, dtype=np.uint16).reshape((h,w)) 

# Save as PNG, and retain 16-bit resolution
Image.fromarray(img).save('result.png')

# Alternative to line above - save as JPEG, but lose 16-bit resolution
Image.fromarray((img>>8).astype(np.uint8)).save('result.jpg') 

两者之间的联系 ipython 它下面使用的python解释器和库是在安装时建立的(而不是每次运行它时动态地建立的),因此我认为如果您删除并重新安装,您的问题将得到解决 伊皮松 ,即:

brew rm ipython
brew install ipython

如果你想了解这个,看看你的 伊皮松 这样的脚本:

more $(which ipython)

直接回答你的问题时, y 轴在第一个 numpy 数组,后跟 x 轴,所以交换索引。

不那么直接,你会发现 for 在Python中循环非常慢,通常最好使用 麻木的 而是矢量化操作。而且,你会发现在 HSV colourspace .

让我们从HSL色轮开始:

假设你想把所有的绿色变成黑色。所以,从维基百科的页面上,绿色对应的色调是120度,这意味着你可以这样做:

#!/usr/local/bin/python3
import numpy as np
from PIL import Image

# Open image and make RGB and HSV versions
RGBim = Image.open("image.png").convert('RGB')
HSVim = RGBim.convert('HSV')

# Make numpy versions
RGBna = np.array(RGBim)
HSVna = np.array(HSVim)

# Extract Hue
H = HSVna[:,:,0]

# Find all green pixels, i.e. where 100 < Hue < 140
lo,hi = 100,140
# Rescale to 0-255, rather than 0-360 because we are using uint8
lo = int((lo * 255) / 360)
hi = int((hi * 255) / 360)
green = np.where((H>lo) & (H<hi))

# Make all green pixels black in original image
RGBna[green] = [0,0,0]

count = green[0].size
print("Pixels matched: {}".format(count))
Image.fromarray(RGBna).save('result.png')

它给出:

这里有一个稍微改进的版本,它保留了alpha/透明度,并匹配红色像素以获得额外的乐趣:

#!/usr/local/bin/python3
import numpy as np
from PIL import Image

# Open image and make RGB and HSV versions
im = Image.open("image.png")

# Save Alpha if present, then remove
if 'A' in im.getbands():
    savedAlpha = im.getchannel('A')
    im = im.convert('RGB')

# Make HSV version
HSVim = im.convert('HSV')

# Make numpy versions
RGBna = np.array(im)
HSVna = np.array(HSVim)

# Extract Hue
H = HSVna[:,:,0]

# Find all red pixels, i.e. where 340 < Hue < 20
lo,hi =  340,20
# Rescale to 0-255, rather than 0-360 because we are using uint8
lo = int((lo * 255) / 360)
hi = int((hi * 255) / 360)
red = np.where((H>lo) | (H<hi))

# Make all red pixels black in original image
RGBna[red] = [0,0,0]

count = red[0].size
print("Pixels matched: {}".format(count))

result=Image.fromarray(RGBna)

# Replace Alpha if originally present
if savedAlpha is not None:
    result.putalpha(savedAlpha)

result.save('result.png')