stretching histogram of an image

I have an image attached with this message and its histogram is also attached. I want to stretch its histogram from 0-255 to 0-65535 and keeping the shape as it is. Please help in this regard.

Thanks in advance.

<<

>>

Answer 1

[Image Analyst]

Simply multiplying the image by 256 will stretch the histogram.

image16 = uint16(image8) * uint16(256);

However, it will leave gaps of 256 between the bins. Is that what you want? If not, then you must cast to double and add noise in the range of 0-255 before scaling:

noiseImage = randsize(image8) - 0.5
image16 = uint16((double(image8) + noiseImage) * 256);

This would give a "solid" histogram with no gaps. The code is untested but I believe it should work.

[try analyst]

Thanks Image Analyst, It is working and I am getting full stretched histogram. But question I want to ask that even the gray levels have been stretched to 16 bit but there is no significant difference in display. In fact, the pixels after stretching must show a more detailed information. How can I do that, if possible?

[Image Analyst]

Of course not. Why should there be? It maps the image range to the display range for each. If you want them to be compared on the same scale, you'll have to specify the scale, like

imshow(image8, [0, 65535]);
imshow(image16, [0, 65535]);

[Thorsten]

You cannot get additional information by stretching the histogram. It's just a different way to store the same data. Like if you have values 1 2 3, you cannot get any more detailed information from 10 20 30. And if you have a true 16 bit image, you need a monitor that can display 16 bit images without truncating them to 8 bit, to see a difference.

[try analyst]

Is the displaying monitor only thing that can improve the quality? I read hist stretching can improve contrast and it results in visualizing more information....any comment?