intensity measure has to be high or low for a good image

Hi friends want to know few things about some parameters

1)(mean) the average intensity of the image has to be high or low for a better image?

2)(standard deviation)the spread of intensities across the image has to be high or low for a better image?

3) (skew) the asymmetry in the image’s intensity distribution has to be high or low for a better image?

4)(entropy) a measure of an image’s smoothness in terms of gray-level values has to be high or low for a better image?

5) (energy) the gray-level values are distributed within the image has to be high or low for a better image?

please guide me ... share some links which describe it has to be high or low ?

[chitresh]

any one there who can answer my question

[David Young]

It's very difficult to answer your question because you do not say what you mean by "better".

[mohammed]

i think you can you use imagej.

http://imagej.nih.gov/ij/

[chitresh]

@David Young : better i mean quality, view, appearance.. if we compare two images then we can say one image is better because its mean, std, skew, entropy and energy is high or low ... hope you get my point

[Image Analyst]

Everyone understood that. We got your point, hence our answers, which remain unchanged despite your replies.

Answer 1

[Image Analyst]

In general, for (1) the mean intensity should be higher for a better image. The more of your available dynamic range you are able to use, the better. For the others it really depends on what is causing the spread in gray levels - noise or the "true" subject. And it really depends on the subject. You can't give a blanket statement that covers all images that the image will be better or worse if those quantities are higher or lower. If you say one thing, you could easily think of an image that is a counterexample to that.

[chitresh]

i want to know all these parameters for the spread in gray level of true reality.. subjects like what, can we study histogram graphs can it tells image is good or bad

[Image Analyst]

For the first part.. sure, go ahead and measure them.

For the second part, see my answer. Like David and I said whether those values indicate a good or bad image depends on what the image is and what those values are supposed to be for a "good" version of that image.

[chitresh]

if we apply image enhancement technique such as

histeq, imsharpen, imhistmatch, decorrstretch then compared to previous image paremeters should be ????

can you link some papers that tells the mean should be high for good quality images

[Image Analyst]

If you apply contrast stretching methods like you listed, the variation/spread in gray levels will be higher. That may or may not produce a more pleasing or more useful image.

I don't know of a paper that says more light is better. It's just common sense so people don't really write papers on that. You can search the web for it and perhaps find links such as "http://www.gcflearnfree.org/digitalphotography/3/print" that say "Generally, the more light you have, the better your photos will turn out. However, you'll also need to think about the quality of the light, as some types of light are more flattering than others"

If you want the technical reason, look up Poisson Noise and photons and photography. You could also look up shot noise and dark current.

[chitresh]

k thanks for the answer but may and may not still confusion not exact point...

k can we interpret image is better or worst through histogram reading if yes then please let me know how