Question

I have a problem in converting the binary to decimal (it seems very lengthy).

%% Read

clear all;
close all;
clc;
I=imread('test.png');
imshow(I);

%% Crop
I2 = imcrop(I);
figure, imshow(I2)
w=size(I2,1);
h=size(I2,2);
%% LBP
for i=2:w-1
    for j=2:h-1
        J0=I2(i,j);
        I3(i-1,j-1)=I2(i-1,j-1)>J0;
        I3(i-1,j)=I2(i-1,j)>J0;
        I3(i-1,j+1)=I2(i-1,j+1)>J0; 
        I3(i,j+1)=I2(i,j+1)>J0;
        I3(i+1,j+1)=I2(i+1,j+1)>J0; 
        I3(i+1,j)=I2(i+1,j)>J0; 
        I3(i+1,j-1)=I2(i+1,j-1)>J0; 
        I3(i,j-1)=I2(i,j-1)>J0;
        LBP(i,j)=I3(i-1,j-1)*2^8+I3(i-1,j)*2^7+I3(i-1,j+1)*2^6+I3(i,j+1)*2^5+I3(i+1,j+1)*2^4+I3(i+1,j)*2^3+I3(i+1,j-1)*2^2+I3(i,j-1)*2^1;
    end
end
figure,imshow(I3)
figure,imhist(LBP)

Is it possible to change this line

LBP(i,j)=I3(i-1,j-1)*2^8+I3(i-1,j)*2^7+I3(i-1,j+1)*2^6+I3(i,j+1)*2^5+I3(i+1,j+1)*2^4+I3(i+1,j)*2^3+I3(i+1,j-1)*2^2+I3(i,j-1)*2^1;

to something shorter?

Answer 1

Option 1:

One way to simplify what you’re doing is to first create a 3-by-3 matrix of scale factors (i.e. powers of two) and initialize it before your loops:

scale = 2.^[8 7 6; 1 -inf 5; 2 3 4];

Then you can replace everything inside your loop with these vectorized operations:

temp = (I2(i-1:i+1,j-1:j+1) > I2(i,j)).*scale;
LBP(i,j) = sum(temp(:));

Option 2:

Alternatively, I believe you can remove both of your loops entirely and replace them with this single call to NLFILTER to get your matrix LBP:

LBP = nlfilter(I2,[3 3],@(x) sum((x(:) > x(5)).*scale(:)));