Python rawpy.imread方法代碼示例

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def load_dng(filename, **kwargs):
    """Loads a dng image file from disk into a float32 Numpy Array (OpenCV view).

    Requires rawpy.

    Args:
        filename (str): Name of pfm image file.
        **kwargs: Extra keyword arguments to pass to `rawpy.postprocess()`.
    """
    import rawpy
    filename = process(filename)
    with rawpy.imread(filename) as raw:
        default_kwargs = dict(gamma=(1,1), no_auto_bright=True, output_bps=16)
        default_kwargs.update(kwargs)
        img = raw.postprocess(**default_kwargs)
    return rgb2bgr(img,-1).astype('float32')

# Accepts hwc - BGR float32 numpy array (cv style)
# TODO: Improve multiple views support
# TODO: Improve speed 

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def imread(filename):
    """Reads an image file from disk into a Numpy Array (OpenCV view).

    Args:
        filename (str): Name of pfm image file.
    """
    filename = process(filename)
    ext = os.path.splitext(filename)[1]
    if ext.lower() == '.pfm':
        return load_pfm(filename)
    elif ext.lower() == '.dng':
        return load_dng(filename)
    else:
        loaded = cv2.imread(filename, flags=cv2.IMREAD_ANYDEPTH + cv2.IMREAD_COLOR)
        if loaded is None:
            raise IOError('Could not read {0}'.format(filename))
        else:
            return loaded 

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def load_image(path):
    if(path[-3:] == 'dng'):
        import rawpy
        with rawpy.imread(path) as raw:
            img = raw.postprocess()
    elif(path[-3:]=='bmp' or path[-3:]=='jpg' or path[-3:]=='png'):
        import cv2
        return cv2.imread(path)[:,:,::-1]
    else:
        img = (255*plt.imread(path)[:,:,:3]).astype('uint8')

    return img 

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def get_visible_raw_image(image_path):
    raw_image = rawpy.imread(image_path).raw_image_visible.copy()
    # raw_image = rawpy.imread(image_path).raw_image.copy()
    return raw_image 

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def load_image(path):
    if(path[-3:] == 'dng'):
        import rawpy
        with rawpy.imread(path) as raw:
            img = raw.postprocess()
        # img = plt.imread(path)
    elif(path[-3:]=='bmp' or path[-3:]=='jpg' or path[-3:]=='png'):
        import cv2
        return cv2.imread(path)[:,:,::-1]
    else:
        img = (255*plt.imread(path)[:,:,:3]).astype('uint8')

    return img 

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def load_image(path):
    if (path[-3:] == 'dng'):
        import rawpy
        with rawpy.imread(path) as raw:
            img = raw.postprocess()
        # img = plt.imread(path)
    elif (path[-3:] == 'bmp' or path[-3:] == 'jpg' or path[-3:] == 'png'):
        import cv2
        return cv2.imread(path)[:, :, ::-1]
    else:
        img = (255 * plt.imread(path)[:, :, :3]).astype('uint8')

    return img 

# 需要導入模塊: import rawpy [as 別名]
# 或者: from rawpy import imread [as 別名]
def apply(model_path, filenames_short, exposure_time_sec):
    import rawpy
    import cv2
    import numpy as np
    from prepare_sony import pack_raw
    import glob
    import tqdm

    if os.path.isfile(filenames_short):
        filenames_short = [filenames_short]
    else:
        filenames_short = glob.glob(filenames_short)

    # assume all images have same shape to avoid multiple loads of the checkpoint
    short_raw = rawpy.imread(filenames_short[0])
    short_uint16 = pack_raw(short_raw)
    short_float = short_uint16.astype(np.float32)
    short_float = np.maximum(short_float - 512, 0) / (16383 - 512)
    H, W, _ = short_float.shape

    predict_func = OfflinePredictor(PredictConfig(
        model=Model(H, W),
        session_init=get_model_loader(model_path),
        input_names=['short_exposure'],
        output_names=['prediction']))

    # apply model to all images
    for filename_short in tqdm.tqdm(filenames_short):
        in_exposure_time = float(filename_short.split('_')[-1].replace('s.ARW', ''))
        factor = exposure_time_sec / in_exposure_time

        short_raw = rawpy.imread(filename_short)
        short_uint16 = pack_raw(short_raw)
        short_float = short_uint16.astype(np.float32)
        short_float = np.maximum(short_float - 512, 0) / (16383 - 512)
        short_float *= factor
        pred = predict_func(short_float[None, ...])[0][0, ...]
        p = np.clip(pred * 255, 0, 255)
        out_fn = 'outputs/{}-{}-prediction.jpg'.format(os.path.basename(filename_short)[:-4], exposure_time_sec)
        cv2.imwrite(out_fn, p[:, :, ::-1])

        naive_uint16 = short_raw.postprocess(use_camera_wb=True, half_size=False, no_auto_bright=True, output_bps=16)
        naive_float = naive_uint16.astype(np.float32) / 65535.0
        # naive_float *= factor
        naive_float *= pred.mean() / naive_float.mean()

        p = np.clip(naive_float * 255, 0, 255)
        out_fn = 'outputs/{}-{}-noise.jpg'.format(os.path.basename(filename_short)[:-4], exposure_time_sec)
        cv2.imwrite(out_fn, p[:, :, ::-1])