Are there standard methods for skin dermatological image preprocessing? And how to achieve efficient methods with minimum operation?
You need in other author. She have another qualification.
I don't undestand you, could you please explaine more.Thank you in advance to reponse me.
If you present an example image, I can recommend proper solutions for preprocessing.
As a general rule, you can perform histogram equalization after implementing a suitable gaussian filter for denoising image..
of course it depends on what you want to achieve. What is the final goal ? is color information important for you , or textural pattern plays more role ?
Hi,Mr.Ömer Faruk , the attached two unhealthy skin images.
In the first image , I need to remove the black region, the dermoscopic gel, the ruler marking, and what is the cause(s) of this effect (black area).
concerning the second image, i need to enhance the image color. I think it is underexposed.
but i don't understand the reason for that?
another question, can i use illumination normalization,color normalization, color quantization, color constancy ,color calibration at the same time.
and in which case i should apply each one?( coud you give me some examples please)
Thank you in advance Mr.Ömer Faruk.
Hi Mr. Fahmi Fahmi, i need to preprocess effeciently the unhealthy skin images with minimum of operation.
i would like to extract unhealthy skin pattern (such as nodules,...).
Can i detect the pattern from color and texture information.
Thank you in advance Mr.Fahmi.
Hello,
In the image performance a lot depend on the way it was taken. If the camera is unpolarized then you should use gel to avoid some unwanted reflections. Try to use that for image acquiring. Further there are several techniques for noise removing, spatial filtering as well as for edge detection. I suggest step approach for feature extraction and then colour operation.
Hello Ms. Khadidja Belattar,
As I understood from your images, you want to eliminate some specific regions from your first image and you want to make some area obvious at second image.
IMO, you can use median filter for denoising black area so you can make those are adaptive to their neighbor pixels. Edge detection may be useful but it has disadvantages on colorful images like feature loss; because you should convert image to gray-scalle to perform edge detection operation properly.
For second image I have tried histogram equalization to every layer(R, G and B) of that color image. The result is like this : http://tinypic.com/r/1zggu2c/8 It gives a tip for this operations to you.
For your last question, there is no theorical background of compound image processing operations. Most of them are based on try-and-see methods. But I do not suggest to apply multiple techniques at once. Because it would not be adaptive for every skin image.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing