Hello,
I would like to test the contrast enhancement method on test dermoscopic skin lesion images but the problem i have not images of poor contrast, so what can i do to create images of bad contrast?
Thank you in advance.
Best Regards
Khadidja BELATTAR
reduce the intensity of image you can make you own code my considering a threshold vale as per your desire you can reduce the intensity above that threshold and increase the intensity below that threshold , contrast is nothing else then intensity difference of object and background.
reduce the intensity of image you can make you own code my considering a threshold vale as per your desire you can reduce the intensity above that threshold and increase the intensity below that threshold , contrast is nothing else then intensity difference of object and background.
Get some image processing software (eg. GIMP) and use the contrast adjustment function. This works in both directions. Be aware that you may have to adjust brightness the opposite direction: until now I haven't met a software where this was not necessary to prevent the image from over-/underexposure effects.
bad contrast can be achieved via https://en.wikipedia.org/wiki/Gamma_correction
Without rejecting other answers, I would consider these three free options:
1- For one image: In GIMP, with tools under Color menu (e.g. directly with Brightness-Contrast...)
2- Batch manipulation of many images in command line with ImageMagick.
http://www.imagemagick.org/script/command-line-options.php#contrast
3- Since it is an easy job, I would write a small code which can interpolate between r, g and b values of each pixel and their gray value L
r_{new} = c * r + (1-c) * L
g_{new} = c * g + (1-c) * L
b_{new} = c * b + (1-c) * L
where, L = 0.3*r + 0.6*g + 0.1*b and c is a constant real number (for all the pixels) between 0 and 1, while c = 1 means no change in contrast. and c = 0 leads to gray scale image. If you mean the contrast in gray scale, then pick a single L_{mean} = mean value of all the pixel intensities L, and in the above interpolation use that mean value. You can adapt L and c based on your needs.
Just for testing of Contrast enhancement, you can refer to the website of the book "Digital Image Processing", by Gonzalez.
For specific image, you can refer to some standard database related to dermoscopic skin disorder as set-up in the below paper. They also told to share the database.
http://users.isr.ist.utl.pt/~jsm/publications/PH2.pdf
As per your requirement, change your contrast of the image by manual stretching method.
by choosing the gray levels of the image using histogram and get the slope.
Use the formula(
Let (a,c) gray level values are mapped to (b,d) using the
s=slope=(d-c)/(b-a)
for example :(5,2) gray levels are mapped to (9,14)
new value=s(i-5)+2; i is the input gray levels between 5 to 9
You can find plenty of images with low contrast in the contrast change subsets from TID2013 and CSIQ databases (See links below). There you can find the original image and 4 to 6 samples of the source image after some contrast decrement.
Additionally you can take a look to the publication 'Computing contrast ratio in images using local content information ', also attached, if you are interested in measuring the quality of the contrast improvement compared with the source image.
http://www.ponomarenko.info/tid2013.htm
http://vision.eng.shizuoka.ac.jp/mod/page/view.php?id=23
Conference Paper Computing contrast ratio in images using local content information
First of all, you have to assign the type of image of image (gray or color) and type of image quality weakness you are intending to handle in your study... The image may hold a limited band of gray levels, or there is unbalance in color bands energies (AC or DC)... or the image covey a sort of noise (with low probability occurrence), or low contrast (whether its type is local or global)...
You can either use Photoshop tools (or some other SW alternatives) to simulate the above mentioned image weaknesses, or writing small codes (using any computer language: e.g., C#, VB, Java, Mat)...
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing