See In The Dark之RAW图片读取-rawpy

前言：处理不同相机的RAW格式图片，并放入卷积运算，输入到see in the dark模型进行夜间识别

1. 保存RAW图片为RGB的JPG格式

import rawpy
import imageio
import numpy as np

# .ARW为索尼Sony相机RAW格式
# .CR2为佳能canon相机RAW格式
raw = rawpy.imread('IMG_0001.CR2')
 
# use_camera_wb 是否执行自动白平衡，如果不执行白平衡，一般图像会偏色
# half_size 是否图像减半
# no_auto_bright 不自动调整亮度
# output_bps bit数据， 8或16
img = raw.postprocess(use_camera_wb=True, half_size=False, no_auto_bright=False, output_bps=16)
 
# img = np.float32(img / (2**16-1)*255.0)
# img = np.asarray(img,np.uint8)
 
imageio.imsave('take.jpg', img)

2. bayer阵列解析

拜耳阵列（Bayer pattern）分为GBRG、GRBG、BGGR、RGGB四种模式：

使用rawpy把不同模式下raw解析为四通道：

import rawpy
import numpy as np
from PIL import Image
import imageio
import exifread
 
def gray_ps(rgb):
    return np.power(np.power(rgb[:, :, 0], 2.2) * 0.2973 + np.power(rgb[:, :, 1], 2.2) * 0.6274
                    + np.power(rgb[:, :, 2], 2.2) * 0.0753, 1 / 2.2) + 1e-7
 
def do_HDR(x, curve_ratio):
    gray_scale = np.expand_dims(gray_ps(x), axis=-1)
    gray_scale_new = np.power(gray_scale, curve_ratio)
    return np.minimum(x * gray_scale_new / gray_scale, 1.0)
 
def adjust_out_matrix(RAW_path, out=None):
    raw = open(RAW_path, 'rb')
    exif_info = exifread.process_file(raw, details=False, strict=True)
    orientation_str = 'EXIF Orientation'
    if exif_info.__contains__('Image Orientation'):
        orientation_str = 'Image Orientation'
    orientation_info = exif_info[orientation_str].printable
    if orientation_info == 'Rotated 180':
        if out is None:
            return True
        else:
            if out.shape[2] == 3:
                out0 = out[:, :, :1]
                out1 = out[:, :, 1:2]
                out2 = out[:, :, 2:3]
                out = np.concatenate((out2, out1, out0), 2)
            elif out.shape[2] == 4:
                out0 = out[:, :, :1]
                out1 = out[:, :, 1:2]
                out2 = out[:, :, 2:3]
                out3 = out[:, :, 3:4]
                out = np.concatenate((out3, out2, out1, out0), 2)
            else:
                raise
            return np.flip(out)
 
    elif orientation_info == 'Horizontal (normal)':
        if out is None:
            return False
        else:
            return out
    else:
        raise
 
# pack Bayer image to 4 channels
def pack_raw(raw_path, white_balance=True, auto_bright=True, HDR=True, save_JPEG=True):
    raw = rawpy.imread(raw_path)
    im = raw.raw_image_visible.astype(np.float32)
 
    # subtract the black level
    # 16383(2^14) is the camera's maximal pixel value, you can get it by "np.max(raw.raw_image)" . Ensure full exposure!
    im = np.maximum(im - raw.black_level_per_channel[0], 0) / (16383 - raw.black_level_per_channel[0])
 
    im = np.expand_dims(im, axis=2)
    H = im.shape[0]
    W = im.shape[1]
 
    if raw.raw_pattern[0, 0] == 0:  # RGGB
        out = np.concatenate((im[0:H:2, 0:W:2, :],
                              im[0:H:2, 1:W:2, :],
                              im[1:H:2, 1:W:2, :],
                              im[1:H:2, 0:W:2, :]), axis=2)
    elif raw.raw_pattern[0, 0] == 2:  # BGGR
        out = np.concatenate((im[1:H:2, 1:W:2, :],
                              im[0:H:2, 1:W:2, :],
                              im[0:H:2, 0:W:2, :],
                              im[1:H:2, 0:W:2, :]), axis=2)
    elif raw.raw_pattern[0, 0] == 1 and raw.raw_pattern[0, 1] == 0:  # GRBG
        out = np.concatenate((im[0:H:2, 1:W:2, :],
                              im[0:H:2, 0:W:2, :],
                              im[1:H:2, 0:W:2, :],
                              im[1:H:2, 1:W:2, :]), axis=2)
    elif raw.raw_pattern[0, 0] == 1 and raw.raw_pattern[0, 1] == 2:  # GBRG
        out = np.concatenate((im[1:H:2, 0:W:2, :],
                              im[0:H:2, 0:W:2, :],
                              im[0:H:2, 1:W:2, :],
                              im[1:H:2, 1:W:2, :]), axis=2)
    # 启用白平衡可以防止图片偏蓝或者偏红
    if white_balance:
        wb = np.array(raw.camera_whitebalance, np.float32)
        wb[3] = wb[1]
        wb = wb / wb[1]
        out = np.minimum(out * wb, 1.0)
    if auto_bright:
        mean_G = (out[:, :, 1].mean() + out[:, :, 3].mean()) / 2.0
        out = np.minimum(out*0.2/mean_G, 1.0)
    out = adjust_out_matrix(raw_path, out)
    # 检测RAW格式解析后的图片是否正常
    if save_JPEG:
        out0 = out[:, :, 0:1]
        out1 = out[:, :, 1:2]
        out2 = out[:, :, 2:3]
        out3 = out[:, :, 3:4]
        out_JPEG = np.concatenate((out0, (out1 + out3) / 2., out2), axis=2)
        if HDR:
            out_JPEG = do_HDR(out_JPEG, 0.35)
        Image.fromarray(np.uint8(out_JPEG * 255)).save('result.jpg')
    return out
 
if __name__ == '__main__':
    raw = rawpy.imread('IMG_0001.CR2')
    np_channel = pack_raw('IMG_0001.CR2', auto_bright=False, HDR=False)
    img = raw.postprocess(use_camera_wb=True, half_size=False, no_auto_bright=True, output_bps=16)
    imageio.imsave('rawpy.jpg', img)

3. 苹果相机iphone 13 pro max 的RAW图片

	raw = rawpy.imread("IMG_0001.DNG")
	im = raw.raw_image_visible.astype(np.float32)  # （h, w, c） = (h, w, 4)
	# 手机竖直拍摄需要加上下面两行，横向拍摄不需要
	if vertical:
		out = np.swapaxes(im,1,0)		# (h w) axis change 20220808
 		out = out[:,::-1,:]   			# Horizontal mirror 20220810

4. 小米12pro 的RAW图片

def pack_raw(raw):
    #pack Bayer image to 4 channels
    im = raw.raw_image_visible.astype(np.float32)
    im = np.maximum(im - 63,0)/ (255 - 63) #subtract the black level

    im = np.expand_dims(im,axis=2)
    img_shape = im.shape
    H = img_shape[0]
    W = img_shape[1]

    out = np.concatenate((im[0:H:2,0:W:2,:],
                       im[0:H:2,1:W:2,:],
                       im[1:H:2,1:W:2,:],
                       im[1:H:2,0:W:2,:]), axis=2)
	# 手机竖直拍摄需要加上下面两行，横向拍摄不需要
	if vertical:
    	out = np.swapaxes(out,1,0)    # (h w) axis change 20220808
    	out = out[:,::-1,:]           # Horizontal mirror 20220810
    print("pack_raw out.shape: ", out.shape)
    return out

raw = rawpy.imread("IMG_0001.DNG")
im = raw.raw_image_visible.astype(np.float32)  # (h, w)
input_images = np.expand_dims(pack_raw(raw),axis=0)