Not sure if I completely follow your question, but you could do so pixel by pixel. If pixel gray level values are not 8 bit, you could typecast them. 

A simpler way is that if A and B denote the two images, do A.*B 

Thanks for your answer Abhinav Jha. I completed the multiplication. But my simulation is running quite long time. This is my code. Can you provide any idea to solve this?

I = imread('peppers.png');

%I = imread('11.jpg');

[m n]=size(I);

%I=im2double(I);

for i=1:m

for j=1:n

%ui = typecast(pi, 'uint8');

b(i,j)= MULTIPLIER_ACCURATE(I(i,j),I(i,j));%just for trail

b(i,j)=1-cos(b(i,j))/2

disp(b);

end

end.

You could find out which portion of your code is taking the most time using the debug routines in Matlab. I have a feeling it is the MULTIPLIER_ACCURATE routine since everything else seems quite straightforward. 

Thank you very much for yours prompt response. I found the solution.