I am working with Content based  skin lesin image Retrieval .These images suffers from black frame artifacts.Attached,you will find an image example.

I remove the back frame by applying gemotric shape(cercle) on the image, and hence the number of pixels is reduced. 

So, if i count the percentage of gray levels in the skin lesion,  i will take the total number of pixels of the original image or the cropped images?

2. is there any effect on the other extracted features (texture features for instance) when i crop the original skin lesion image?

3. What is the difference betwwen the relative and absolute image features?

4. Is there better solution to remove the black frame without cropping the image? 

Thank You

Khadidja BELATTAR

Best Regards.

More Khadidja Belattar's questions See All
Similar questions and discussions