Interpolation (of Matlab implementation) of pixels in the grayscale image

Common interpolation pixel grayscale method: nearest neighbor method, the bilinear interpolation, cubic interpolation method

%% pixel interpolation gray-scale 
factor = 0.75 ;% scaling ratio 
U = 0.6 ; V = 0.7 ; 
itp1 = uint8 (zeros (ceil (H / factor), ceil (W / factor))); 
ITP2 = uint8 (zeros ( ceil (H / U), ceil (W / V))); 
itp3 = uint8 (zeros (ceil (H / U), ceil (W / V)));

J3 following is a grayscale image to be interpolated, the interpolation image ITP

1. nearest neighbors method

         Seeking to be in the neighborhood pixels of pixels, the pixels required to be assigned to the nearest neighbor pixel gray this point.

%最近邻元法
 for i = 1:h/factor
    for j = 1:w/factor
        itp1(i,j) = J3(floor(i*factor+0.5),floor(j*factor+0.5));
    end
 end

2. the bilinear interpolation

%双线性内插法
for i = 2:h/u-1
    for j = 2:w/v-1
        itp2(i,j) = (1-(i*u-floor(i*u)))*(1-(j*v-floor(j*v)))*J3(floor(i*u),floor(j*v))...
                   +(1-(i*u-floor(i*u)))*(j*v-floor(j*v))*J3(floor(i*u),floor(j*v)+1)...
                   +(i*u-floor(i*u))*(1-(j*v-floor(j*v)))*J3(floor(i*u)+1,floor(j*v))...
                   +(i*u-floor(i*u))*(j*v-floor(j*v))*J3(floor(i*u)+1,floor(j*v)+1);         

    end
end

3. tertiary within 插法

function [f] = S(x)
if (0<=abs(x))&&(abs(x)<1)
    f = 1-2*(abs(x))^2+(abs(x))^3;
elseif (1<=abs(x))&&(abs(x)<2)
    f = 4-8*abs(x)+5*(abs(x))^2-(abs(x))^3;
else
    f = 0;
end
end


%三次内插法
f = double(J3);
u = 0.6;v = 0.7;
for i = ceil(2/u):h/u-2
    for j = ceil(2/v):w/v-2
       A = [S(1+(j*v-floor(j*v))) S((j*v-floor(j*v))) S(1-(j*v-floor(j*v))) S(2-(j*v-floor(j*v)))];
       C = [S(1+(i*u-floor(i*u)));S((i*u-floor(i*u)));S(1-(i*u-floor(i*u)));S(2-(i*u-floor(i*u)))];
       i1 = floor(i*u); j1 = floor(j*v);
       B = [f(i1-1,j1-1) f(i1-1,j1) f(i1-1,j1+1) f(i1-1,j1+2);
            f(i1,j1-1) f(i1,j1) f(i1,j1+1) f(i1,j1+2);
            f(i1+1,j1-1) f(i1+1,j1) f(i1+1,j1+1) f(i1+1,j1+2);
            f(i1+2,j1-1) f(i1+2,j1) f(i1+2,j1+1) f(i1+2,j1+2);];
        itp3(i,j) = A*B*C;
    end
end

 

上述的内插法处理的结果出现了黑边，可以通过将黑边区域赋予原始图像的对应区域的像素灰度值。

可通过以下循环实现。

%去黑边
    [m,n] = size(J3);%原始图像的行列数

    for i = 1:floor(1/u) 
        for j = 1:floor(n/v)
          itp(i,j) = J3(ceil(i*u),ceil(j*v));
          itp(floor(m/u),j) = J3(m,ceil(j*v));
        end
    end
    for j = 1:floor(1/v)
        for i = 1:floor(m/u)
           itp(i,j) = J3(ceil(i*u),ceil(j*v));
           itp(i,floor(n/v)) = J3(ceil(i*u),n);
        end
    end