2
$\begingroup$

I am a beginner in digital image processing, recently I read a paper on color image contrast enhancement, In that paper authors have implemented their method in HSV color model. In my work I have implemented that paper in MATLAB using RGB color model. I just want to know that whether it will make any difference or not?

$\endgroup$
  • $\begingroup$ I tried using HSV model for retinal (fundus)images but the background and blood vessels are showing almost showing same values(as the background is also red in color)...how can I suppress background or please suggest me the suitable color plane. Thanks in advance! $\endgroup$ – Sarika Patil May 4 '18 at 14:06
  • $\begingroup$ @SarikaPatil Welcome to SE.DSP! Please ask a new question. Don't ask a question as an answer to an old question. $\endgroup$ – Peter K. May 4 '18 at 18:04
4
$\begingroup$

HSV is one of the color space transforms that separate color from intensity. Depending on your application (my understanding is contrast enhancement is somehow related to image segmentation in your OP), there are several advantages of HSV:

  1. Histogram equalization of a color image is suggested only on the intensity component;

  2. The Saturation component is regarded as invariant in shadow (only intensity changes), so processing the image in HSV space is better suit for shadow removing. This post that tried to remove the shadow from an image is for your reference;

  3. The Hue component makes the algorithm better immune and thus more robust to lighting variations. This feature makes it better fit in some medical applications such as the palm skin segmentation (under normal RGB space it is not easy to thresholding out the hand palm).

  4. In color enhancement that you mentioned, if the colors in your image are not so strong, it would be quite difficult to detect colored objects. Adding some arbitrary value in the Image would not Help, since it would increase the brightness but not the contrast. After the conversion to HSV space you may increase the Saturation then convert it back to the RGB color space. The colors of interest may appear much better in the image which would facilitate the object recognition nicely. You can have a try by this means to adjust your image contrast.

You may also want to check this post for more discussions.

$\endgroup$
1
$\begingroup$

RGB module has three dependent layers R, G, B any change in the lightness and brightness reflects on the three colors layers (high correlation between the three layers). unlike other modules i.e, HSV, YIQ which separate the chromatic (color) in one or more independent layer as possible, a comparison between the RGB and other modules briefly was introduced by cheng et al. lightness represents a challenge in the image processing field not only on segmentation but also it has been observed the the affect of illumination variation is more the changes between the individual on pattern recognition. in 2D image, to obtain effective results with low computational cost you should treat the luminance as separated layer. on the other hand, HSV plots three layers: (V) brightness, (H) hue, and (S) saturation. While brightness embodies the achromatic layer, hue represents the dominant color, and saturation the amount of white mixed with the hue (hue and saturation are chromatic layers)

$\endgroup$
0
$\begingroup$

Yes, it make large difference. Why? Because HSV and RGB are much difference color space. RGB is Red, Green and Blue, normal color space. HSV is Hue, Saturation and Value (read more here). In your algorithm ( or paper) take a look in which channel of image you should work. Sometimes you need only Value (for example histogram equalization), sometimes Saturation. In you paper should be precisly be said which channel they are using.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.