how to find existence of noise and edges in a color image using gradient?

matlab code to find existence of noise and edges in a color image using gradient?

Answer 1

[Image Analyst]

Try subtracting the image from your noise-free version to get the noise. Then, have you tried to use imgradient? Or you could try a local delta E (color difference) as given in my demo code: http://www.mathworks.com/matlabcentral/answers/73741#comment_145951

[ARUN SAI]

i have tried this code for finding the existence of noise and edges in a color image named as mandi.tif which is cfa raw image in the matlab library.

 for i=2:1:m-1

     for j=2:1:n-1

% horizontal internal gradient

A1=abs(r(i-1,j-1)-r(i+1,j+1));

% vertical internal gradient

B1=abs(r(i-1,j+1)-r(i+1,j-1));

    end

 end

 for i=3:1:m-2

    for j=3:1:n-2

% horizontal external gradient

      A2=abs(2*g(i,j)-g(i-1,j-2)-g(i+1,j+2));

% vertical external gradient

      B2=abs(2*g(i,j)-g(i-2,j-1)-g(i+2,j+1)); 

    end
 end

% to find existence of edge or influence of the noise TH (enumeration variable) is used for noise that in the up and down or in any point of g(i,j)

 for k=2:2:m-1

     for l=3:2:n-1

        x=[k-1,k+1];

        y=[l-1,l+1];

%up says that g(i,j-1) is the noise

 up=(A1<B1)&(eq(A2,B2))&eq(max(abs(2*r(k,y)-r(k-1,l)-r(k+1,l))),l-1);

%down says that g(i,j+1) is the noise

 down=(A1<B1)&(eq(A2,B2))&eq(max(abs(2*r(k,y)-r(k-1,l)-r(k+1,l))),l+1);

%left says that g(i-1,j) is the noise

 left=(A1>B1)&(eq(A2,B2))&eq(max(abs(2*r(x,l)-r(k,l-1)-r(k,l+1))),k-1);

%right says that g(i+1,j) is the noise

 right=(A1>B1)&(eq(A2,B2))&eq(max(abs(2*r(x,l)-r(k,l-1)-r(k,l+1))),k+1);

*%no says that there is no noise and also no edge *

 no=(eq(A1,B1))&(eq(A2,B2));

% level says that there is a edge in vertical direction

 level=(A1>B1)&(A2>B2);

% erect says that there is a edge in horizontal direction

 erect=(A1>B1)&(A2<B2);

    end

 end

 p=[1 2 3 4 5 6 7]

 TH=[up down left right no level erect]

 p_TH=p(TH==1)

 for i=2:1:m-1

    for j=2:1:n-1

% switches according to the TH value switch p_TH

    case 1
        r(i,j)=(r(i-1,j)+r(i+1,j)+r(i,j+1))/3;
        r(i,j-1)=r(i,j);

    case 2
        r(i,j)=(r(i-1,j)+r(i+1,j)+r(i,j-1))/3;
        r(i,j+1)=r(i,j);

    case 3
        r(i,j)=(r(i+1,j)+r(i,j-1)+r(i,j+1))/3;
        r(i-1,j)=r(i,j);

    case 4
        r(i,j)=(r(i-1,j)+r(i,j-1)+r(i,j+1))/3;
        r(i+1,j)=r(i,j);

    case 5
        r(i,j)=(r(i-1,j)+r(i+1,j)+r(i,j-1)+r(i,j+1))/4;

    case 6
        r(i,j)=(r(i-1,j)+r(i+1,j))/2;

    case 7
        r(i,j)=(r(i,j-1)+r(i,j+1))/2;

 end

    end

 end

 s=r+g+b;

 figure,imshow(s);

%signal to noise ratio calculation of image after removal of noise

 m3=mean2(s);

 SD3=mean2(stdfilt(s));

 snr3=(m3/SD3)