Naveen , open the image file using hex editor and you will find the binary equivalent of each pixel value . If you are using Linux then octal dump (od)command should do.

Best case: open it in C/C++  and read each unsigned char  and check if pixel values are zeros..

Why not. Especially considering your other question about disparity map of binary images: what else could it contain than 1s and 0s ?

Regards

Thank you all.  

I am attaching a file. Please find the attachment. It contains the results of the experiment.

Ah, I can see what you are trying: find matching points via squared differences.

This does not work ! As dispMap[] is already indicating. Thus further discussion of your algorithms is currently not considered.

How depth information is "normally" retrieved:

Locate "prominent" image elements in both images indpendently: corners, edges, borders and alike.

Find the corresponding elements in both images. The difference in relative coordinates of corresponding elements (pair-wise) gives the disparity (?) - better (relative) displacement - of each element pair.

Perform triangulation for each corresponding pair.

I suspect that you did not yet realize that depth information cannot be calculated for arbitrary image elements: this is the same as with human sight: seeing a grey plane without any edges does not allow for distance assessment. It's the edges, corners etc. that allows for depth assessment.

Hope this helps a bit

Since you already are using the DoF , based on lens parameters , 

http://www.dofmaster.com/dofjs.html

I do not want to specify the camera  parameters , the pixel (in micron of the CCD) also plays a role (as a correction factor if you may ) .

It may seem a bit tedious but the numerical aspect can be readjusted in the calculation. 

Hi Naveenkumar,

in a "true" disparity map of two stereo cameras with parallel axes there shouldn't occur any zeros, because a zero disparity corresponds to an infinite distance. In practice, however, zero values in the disparity map can result from finite accuracy / rounding effects or indicate that no match could be found for the corresponding image point. Zero disparity values can also be found regularly in a setting with converging (non-parallel) stereo cameras.

Best regards, Ralf