Image processing 4

Example 1
Image histogram and the appropriate threshold choice
Basing on the histogram characteristics, two different thresholds were chosen (120
and 180). Please, assess the results.

L1=imread(

'portret.jpg'

)

figure
imshow(L1)
set(gcf,

'Color'

,([1 1 1]));figure;

grid;
imhist(L1);
ylabel(

'Ilość pikseli'

,

'FontSize'

,15,

'FontName'

,

'Arial CE'

)

L2a=L1>120;
figure;
imshow(L2a);
L2b=L1>180;
figure;
imshow(L2b)

Example 2
Threshold calculation using Otsu algorithm

L1=imread(

'portret.jpg'

)

figure
imshow(L1)
set(gcf,

'Color'

,([1 1 1]));figure;

grid;
imhist(L1);
ylabel(

'Pixels number

,

'FontSize'

,15,

'FontName'

,

'Arial CE'

)

level = graythresh(L1);
L2 = im2bw(L1,level);
imshow(L2)

Exercise 1
Take any image and, after its histogram analyses, try to choose the best threshold. Is
this possible for every image?
Then, use Otsu algorithm.

Exercise 2
Basing on image „cells.bmp” histogram observation, try to choose two thresholds
(below and above) in order to show cells contours. This exercise should be done trial-
and-error method.

Example 3
Two grayscale images (with the same dimensions) adding.
In the first resulting image brightness is normalized, the second one – is without
normalization.

[L1a,map1a] = imread(

'gory.bmp'

);

L1a=double(ind2gray(L1a,map1a))/255;
figure; imshow(L1a);
[L1b,map1b] = imread(

'drzewo.bmp'

);

L1b=double(ind2gray(L1b,map1b))/255;
figure;imshow(L1b)
L2a=L1a+L1b;
figure; imshow(mat2gray(L2a))
figure; imshow(uint8(round(L2a*255)))

Example 4
Two images adding that have different dimensions

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1a=ind2gray(L1a,map1a);
figure; imshow(L1a);
L1a=L1a(201:650,101:700); % appropriate part cutting
figure; imshow(L1a);
[L1b,map1b] = imread(

'drzewo.bmp'

);

L1b=ind2gray(L1b,map1b);
figure;imshow(L1b)
L2a=L1a+L1b;
figure; imshow(mat2gray(L2a))

Caution:
If the images are of different dimensions, it is necessary to cut the appropriate
fragment from the bigger one. In the other case, the operation cannot be done,
because matrix dimensions must agree.

Example 5

Method 1
Two images mixing with different coefficients

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1a=ind2gray(L1a,map1a);
L1a=L1a(201:650,101:700);
figure; imshow(L1a);
[L1b,map1b] = imread(

'drzewo.bmp'

);

L1b=ind2gray(L1b,map1b);
figure;imshow(L1b)
L2a=imlincomb(0.3,L1a,0.7,L1b); % coefficients total has to be equal 1
figure; imshow(L2a)

Method 2

Two images mixing with different coefficients (changing every 2 seconds)

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1a=ind2gray(L1a,map1a);
L1a=L1a(201:650,101:700);
figure; imshow(L1a);
[L1b,map1b] = imread(

'drzewo.bmp'

);

L1b=ind2gray(L1b,map1b);
figure;imshow(L1b);
figure

for

i=0:0.1:1

L2=(double(L1a)*i+double(L1b)*(1-i))/255;
imshow(L2); title([

'L1a'

,mat2str(i*100),

'%'

,

' + L1b'

,mat2str((1-

i)*100),

'%'

])

pause(2)

end

Example 6
"Imadd" function adds the constant value to the image or adds two images.
Adding constant to every pixel

L1 = imread(

'portret.jpg'

);

figure; imshow(L1)
L2 = imadd(L1,80);
figure; imshow(L2)

Exercise 3
Add any two images using “imadd” function. Be careful, regarding matrices
dimensions.

Example 7
Grayscale images “subtraction” (a hidden image)

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1a=ind2gray(L1a,map1a);
L1a=L1a(201:650,101:700)>128;
figure; imshow(L1a);
[L1b,map1b] = imread(

'drzewo.bmp'

);

L1b=ind2gray(L1b,map1b);
figure;imshow(L1b);
L1c=0.05*double(L1a)/255+0.95*double(L1b)/255;
figure; imshow(L1c);
L2=mat2gray(L1c-0.95*double(L1b)/255)
figure; imshow(L2)

Exercise 4
Please, conduct image subtraction using “imabsdiff” function. Change the
images sequence and see if there is any difference in the resulting image.

Exercise 5
Now use “imsubtract” function for the same operation. Has the image
sequence any consequence in the resulting image?

Example 8
Image multiplication
Multiplication and dividing of different images can be used in the noise reduction.
In landscape photography distant objects lack in sharpness, due to the fog that is
visible in the blue colour channel. Channels multiplication (red and green) using
“immultiply” allows to distinct greenery.

L1=imread(

'gory.jpg'

)

figure;imshow(L1)
L1r=double(L1(:,:,1))/255
figure;imshow(L1r)
L1g=double(L1(:,:,2))/255
figure;imshow(L1g)
L1b=double(L1(:,:,3))/255
figure;imshow(L1b)
L2=immultiply(L1g,L1r)
figure;imshow(L2,[])

Exercise 6
Divide the green channel by the red one using "imdivide".

Example 9
Minimum and maximum (from two images) calculation for every pixel at the
same location

L1=imread(

'gory.jpg'

)

figure;imshow(L1)
L1r=double(L1(:,:,1))/255
figure;imshow(L1r)
L1g=double(L1(:,:,2))/255
figure;imshow(L1g)
L1b=double(L1(:,:,3))/255
figure;imshow(L1b)
L2a=max(L1g,L1b)
L2b=min(L1g,L1b)
figure
imshow(L2a)
figure
imshow(L2b)

Example 10
Method 1
Logical operation NOT for the binary and grayscale image

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1=ind2gray(L1a,map1a);
L1a=L1>128;
figure; imshow(L1a);
L2a=~(L1a);
figure;imshow(L2a);
figure;imshow(L1);
figure;imshow(bitcmp(L1,8));

Method 2
Image completion obtained by using "imcomplement" function

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1=ind2gray(L1a,map1a);
figure; imshow(L1);
L2=imcomplement(L1)
figure;imshow(L2);

Method 3
An identical resulting image obtained by colour palette “grayscale” reversion

[L1a,map1a] = imread(

'Beaux.bmp'

);

L1=ind2gray(L1a,map1a);
figure(

'Color'

,

'w'

); imshow(L1);

colormap(gray(256));
colorbar;
figure(

'Color'

,

'w'

); imshow(L1);

colormap(flipud(gray(256)));
colorbar;

Example 11
Method 1
Binary images conjunction

L1a=imread(

'wykr_10.tif'

)

figure;imshow(L1a)
L1b=imread(

'wykr_14.tif'

)

figure;imshow(L1b)
L2=L1a&L1b
figure; imshow(L2)

Method 2

L1a=imread(

'wykr_10.tif'

)

figure;imshow(L1a)
L1b=imread(

'wykr_14.tif'

)

figure;imshow(L1b)
L2=bitand(L1a,L1b)
figure; imshow(L2)

Exercise 7
Basing on the example above, try to conduct OR, XOR and SUB operations.

Example 12
Grayscale images conjunction

L1=imread(

'gory.jpg'

)

figure;imshow(L1)
L1r=L1(:,:,1)
figure;imshow(L1r)
L1g=L1(:,:,2)
figure;imshow(L1g)
L1b=L1(:,:,3)
figure;imshow(L1b)
L2=bitand(L1r,L1g)
figure; imshow(L2)

Example 13
Image projection on the surface of cylinder

[x,y,z] = cylinder;
L=imread(

'gory.jpg'

);

warp(x,y,-z,L);

... or on the solid that looks like a vase

t=2/3:1/6:2*pi;
y=sin(t)+3
[x,y,z] = cylinder(y);
L=imread(

'gory.jpg'

);

warp(x,y,-z,L);

Remove “-“ sign for “z” coordinate and see the difference.