Vectorisation for nested for loops in Piecewise Linear Transformation

Hi everyone,

I'm trying to remove the for loops in this piecewise linear transformation function and have tried using vectorisation but can't wrap my head around how the code would look.

for i = 1:size(original,1)
    for j = 1:size(original,2)
        if (original(i,j) < x1)
            transformed(i,j) = m1 * original(i,j);
        elseif (original(i,j) >= x1 && original(i,j) < x2)
            transformed(i,j) = m2 * (original(i,j) - x1) + y1;
        else
            transformed(i,j) = m3 * (original(i,j) - x2) + y2;
        end
    end
end

Is there any way I can do a piecewise linear transformation (contrast stretching) without the nested for loop to speed up the execution?

Answer 1

[Voss]

transformed = original;

idx = original < x1;
transformed(idx) = m1 * original(idx);

idx = original >= x1 & original < x2;
transformed(idx) = m2 * (original(idx) - x1) + y1;

idx = original >= x2;
transformed(idx) = m3 * (original(idx) - x2) + y2;

[Sophia]

Thank you so much! It makes a lot of sense now :)