how to find object map from saliency map of an image?????

The codings that i used to find the saliency map of an image is given below:

%%Read image from file 
inImg = im2double(rgb2gray(imread('13.jpg')));
inImg = imresize(inImg, [64, 64], 'bilinear');
subplot(1,2,1);
imshow(inImg);title('Original Image');

%%Spectral Residual
FFT = fft2(inImg); 
LogAmplitude = log(abs(FFT));
Phase = angle(FFT);
SpectralResidual = LogAmplitude - imfilter(LogAmplitude, fspecial('average', 3), 'replicate'); 
saliencyMap = abs(ifft2(exp(SpectralResidual + i*Phase))).^2;

%%After Effect
saliencyMap = mat2gray(imfilter(saliencyMap, fspecial('disk', 3)));
subplot(1,2,2);
imshow(saliencyMap, []);title('Saliency Map');

After getting the saliency map, i need to find the object map of the given input image which i am not able to. Please help!!!!!!!!!!

[Image Analyst]

What is the object map?

[Sansri Basu]

object map refers to the proto objects in the given images.

Given S(x) of an image, the object map O(x) is obtained: O(x) = 1 if S(x) > threshold, 0 otherwise.

Empirically, we set threshold = E(S(x)) × 3, where E(S(x)) is the average intensity of the saliency map. The selection of threshold is a trade-off problem between false alarm and neglect of objects.

We need to use the above theory for the matlab code.

I have attached two images that shows the saliency map as well as the object map of the given input image.

Please help!!!!!

[Prof M S Prasad]

its an old problem . take histogram then select max value as threshold. then 1 sigma value. one of them would produce good result.

[Prof M S Prasad]

hope you got the result.

threshold selection is based on Histogram. analyse one or we can set it 2 sigma.

i think it will give results.

Answer 1

[Image Analyst]

It looks like it was just thresholded:

objectMap = saliencyMap > someThresholdValue;

I don't know what someThresholdValue might be. It's possible it might change on an image by image basis.

[Sansri Basu]

I tried using this code:

t=graythresh(saliencyMap); bw=im2bw(saliencyMap,t); imshow(bw);

But there was a blank output

[Image Analyst]

Maybe that didn't pick the correct threshold. Try it the way I suggested. You can use the interactive thresholding app in my File Exchange if you want. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862