由於人眼對顏色不敏感，而對光亮通道更加敏感。因此在超解析度任務中，我們通常需要將RGB通道轉換為Ycbcr通道。在Python的程式碼實現中，我發現opencv的RGB轉Ycbcr的計算方式和Matlab的實現方式有些不同，而NTIRE的評估往往是在matlab平臺的。因此，這裡需要注意。

Python RGB轉Ycbcr通道

對於Set5中的baby影象

Code:

img = cv2.imread(imgpath)
img = cv2.cvtColor(img, cv2.COLOR_BGR2YCR_CB)
img_y = img[:,:,0]
 

Result:

array([[253, 253, 253, ..., 254, 254, 254],
       [253, 253, 253, ..., 254, 254, 254],
       [253, 253, 253, ..., 254, 254, 254],
       ...,
       [ 62,  70,  72, ...,  67,  67,  67],
       [ 54,  58,  59, ...,  69,  68,  68],
       [ 49,  52,  53, ...,  70,  70,  69]], dtype=uint8)
 

實驗原理:

參考連結：https://docs.opencv.org/3.0.0/de/d25/imgproc_color_conversions.html

Matlab RGB轉Ycbcr通道

Code:

im  = imread(imgpath);
im = rgb2ycbcr(im);
im = im(:, :, 1);

Result:

Matlab實現方式:

function ycbcr = rgb2ycbcr(varargin)
%RGB2YCBCR Convert RGB color values to YCbCr color space.
%   YCBCRMAP =
 
 RGB2YCBCR(MAP) converts the RGB values in MAP to the YCBCR
%   color space. MAP must be a M-by-3 array. YCBCRMAP is a M-by-3 matrix
%   that contains the YCBCR luminance (Y) and chrominance (Cb and Cr) color
%   values as columns.  Each row represents the equivalent color to the
%   corresponding row in the RGB colormap.
%
%   YCBCR = RGB2YCBCR(RGB) converts the truecolor image RGB to the
%   equivalent image in the YCBCR color space. RGB must be a M-by-N-by-3
%   array.
%
%   If the input is uint8, then YCBCR is uint8 where Y is in the range [16
%   235], and Cb and Cr are in the range [16 240].  If the input is a double,
%   then Y is in the range [16/255 235/255] and Cb and Cr are in the range
%   [16/255 240/255].  If the input is uint16, then Y is in the range [4112
%   60395] and Cb and Cr are in the range [4112 61680].
%
%   Class Support
%   -------------
%   If the input is an RGB image, it can be uint8, uint16, or double. If the
%   input is a colormap, then it must be double. The output has the same class
%   as the input.
%
%   Examples
%   --------
%   Convert RGB image to YCbCr.
%
%      RGB = imread('board.tif');
%      YCBCR = rgb2ycbcr(RGB);
%
%   Convert RGB color space to YCbCr.
%
%      map = jet(256);
%      newmap = rgb2ycbcr(map);
%
%   See also NTSC2RGB, RGB2NTSC, YCBCR2RGB.

%   Copyright 1993-2010 The MathWorks, Inc.  

%   References: 
%     C.A. Poynton, "A Technical Introduction to Digital Video", John Wiley
%     & Sons, Inc., 1996, p. 175
% 
%     Rec. ITU-R BT.601-5, "STUDIO ENCODING PARAMETERS OF DIGITAL TELEVISION
%     FOR STANDARD 4:3 AND WIDE-SCREEN 16:9 ASPECT RATIOS",
%     (1982-1986-1990-1992-1994-1995), Section 3.5.

rgb = parse_inputs(varargin{:});

%initialize variables
isColormap = false;

%must reshape colormap to be m x n x 3 for transformation
if (ndims(rgb) == 2)
  %colormap
  isColormap=true;
  colors = size(rgb,1);
  rgb = reshape(rgb, [colors 1 3]);
end

% This matrix comes from a formula in Poynton's, "Introduction to
% Digital Video" (p. 176, equations 9.6). 

% T is from equation 9.6: ycbcr = origT * rgb + origOffset;
origT = [65.481 128.553 24.966;...
     -37.797 -74.203 112; ...
     112 -93.786 -18.214];
origOffset = [16;128;128];

% The formula ycbcr = origT * rgb + origOffset, converts a RGB image in the range
% [0 1] to a YCbCr image where Y is in the range [16 235], and Cb and Cr
% are in that range [16 240]. For each class type (double,uint8,
% uint16), we must calculate scaling factors for origT and origOffset so that
% the input image is scaled between 0 and 1, and so that the output image is
% in the range of the respective class type.

scaleFactor.double.T = 1/255;      % scale output so in range [0 1].
scaleFactor.double.offset = 1/255; % scale output so in range [0 1].
scaleFactor.uint8.T = 1/255;       % scale input so in range [0 1].
scaleFactor.uint8.offset = 1;      % output is already in range [0 255].
scaleFactor.uint16.T = 257/65535;  % scale input so it is in range [0 1]  
                                   % and scale output so it is in range 
                                   % [0 65535] (255*257 = 65535).
scaleFactor.uint16.offset = 257;   % scale output so it is in range [0 65535].

% The formula ycbcr = origT*rgb + origOffset is rewritten as 
% ycbcr = scaleFactorForT * origT * rgb + scaleFactorForOffset*origOffset.  
% To use imlincomb, we rewrite the formula as ycbcr = T * rgb + offset, where T and
% offset are defined below.
classIn = class(rgb);
T = scaleFactor.(classIn).T * origT;
offset = scaleFactor.(classIn).offset * origOffset;

%initialize output
ycbcr = zeros(size(rgb),classIn);

for p = 1:3
  ycbcr(:,:,p) = imlincomb(T(p,1),rgb(:,:,1),T(p,2),rgb(:,:,2), ...
                         T(p,3),rgb(:,:,3),offset(p));
end  

if isColormap
  ycbcr = reshape(ycbcr, [colors 3 1]);
end

%%%
%Parse Inputs
%%%
function X = parse_inputs(varargin)

narginchk(1,1);
X = varargin{1};

if ndims(X)==2
  % For backward compatibility, this function handles uint8 and uint16
  % colormaps. This usage will be removed in a future release.

  validateattributes(X,{'uint8','uint16','double'},{'nonempty'},mfilename,'MAP',1);
  if (size(X,2) ~=3 || size(X,1) < 1)
    error(message('images:rgb2ycbcr:invalidSizeForColormap'))
  end
  if ~isa(X,'double')
    warning(message('images:rgb2ycbcr:notAValidColormap'))
    X = im2double(X);
  end

elseif ndims(X)==3
  validateattributes(X,{'uint8','uint16','double'},{},mfilename,'RGB',1);
  if (size(X,3) ~=3)
    error(message('images:rgb2ycbcr:invalidTruecolorImage'))
  end
else
  error(message('images:rgb2ycbcr:invalidInputSize'))
end

實驗可發現兩種實現方式的結果存在著不同， 這是因為兩者的內部實現原理不同。這裡提供一個與Matlab的Ycbcr空間轉換類似的函式：

def rgb2ycbcr(img, only_y=True):
    '''same as matlab rgb2ycbcr
    only_y: only return Y channel
    Input:
        uint8, [0, 255]
        float, [0, 1]
    '''
    in_img_type = img.dtype
    img.astype(np.float32)
    if in_img_type != np.uint8:
        img *= 255.
    # convert
    if only_y:
        rlt = np.dot(img, [65.481, 128.553, 24.966]) / 255.0 + 16.0
    else:
        rlt = np.matmul(img, [[65.481, -37.797, 112.0], [128.553, -74.203, -93.786],
                              [24.966, 112.0, -18.214]]) / 255.0 + [16, 128, 128]
    if in_img_type == np.uint8:
        rlt = rlt.round()
    else:
        rlt /= 255.
    return rlt.astype(in_img_type)