在Matlab中将图像转换为灰度和十进制范数

Question

我想将图像从其原始RGB比例转换为灰度和十粒瘤。但是，我尝试的每一次尝试都会给我一些较暗的图像，但并不能真正显示出灰度。这是我正在运行的内容

My attempt  for grayscale

至于申命记，我不知道该怎么做。任何帮助，将不胜感激。

Answer 1

您的错误是缩放

whe = imread('CoWheel.png');

whesi = size(whe);


red = [];
gree = [];
blue = [];

for i = 1:whesi(1)
    for j = 1:whesi(2)
        red(i,j) = whe(i,j,1);
        gree(i,j) = whe(i,j,2);
        blue(i,j) = whe(i,j,3);
    end
end

gray = whe;
for i = 1:whesi(1)
    for j = 1:whesi(2)

        gray(i,j,1) = gray(i,j,1)*.2989;
        gray(i,j,2) = gray(i,j,2)*.5870;
        gray(i,j,3) = gray(i,j,3)*.1141;

    end

，gray(i,j,1)，gray(i,j,2)。您需要缩放和求和RGB元素：gray(i,j,3)。

这里是更正的代码示例：

gray = R*.2989 + G*.5870 + B*.1141

结果：

whe = imread('CoWheel.png');
figure;imshow(whe);

whesi = size(whe);


%You don'y use red, green and blue in your conversion code...
% red = [];
% gree = [];
% blue = [];
% 
% for i = 1:whesi(1)
%     for j = 1:whesi(2)
%         red(i,j) = whe(i,j,1);
%         gree(i,j) = whe(i,j,2);
%         blue(i,j) = whe(i,j,3);
%     end
% end

%Convert from uint8 to double (for performing the computations in double - not in uint8 integers).
whe = double(whe);

%gray = whe;

%The Grayscale image has only one color channel (not 3 like RGB)
gray = zeros(whesi(1), whesi(2));

for i = 1:whesi(1)
    for j = 1:whesi(2)
        gray(i,j) = whe(i,j,1)*.2989 + whe(i,j,2)*.5870 + whe(i,j,2)*.1141;
    end
end

%Convert from double to uint8
gray = uint8(gray);

figure;imshow(gray)

---

您也可以使用：

rgb2gray

---

rgb2gray不是感知准确的。如果需要准确的结果，则需要反转伽玛，计算Y（XYZ颜色空间的Y，然后执行伽玛）。

%Using MATLAB rgb2gray built in function:
whe = imread('CoWheel.png');
gray = rgb2gray(whe);

结果：

%Shortcut: convert RGB to LAB, zero A and B, and convert to LAB to RGB
RGB = whe;
LAB = rgb2lab(RGB); %Convert RGB to LAB
LAB(:,:,2) = 0; %Zero A channel
LAB(:,:,3) = 0; %Zero B channel
gray = lab2rgb(LAB); %The result is Grayscale image in RGB format (where R=G=B).
gray = im2uint8(gray); %Convert from range [0, 1] to uint8 [0, 255].
figure;imshow(gray)

这里是RGB图像，第一个灰度结果和可感知的准确结果：

关于申命记，它需要更多的研究...