I am trying to create a machine learning model to classify cervical cancer images. These are 2-D RGB images. Is there an image processing library designed to preprocess these images and pin point the location of a predefined feature in the image itself.
OpenCV, in combination with C++ or Python could be an option.
http://docs.opencv.org/2.4/doc/tutorials/imgproc/table_of_content_imgproc/table_of_content_imgproc.html
By preprocessing you usually remove information.
What can be done is to shift information from
the overall image to a local place. What that
means depends on you specific problem. If ethics
does not forbid it, could you please post a
sample image, or one painted by yourself?
Regards,
Joachim
I do not have much experience in medical images, but I think what you need from RGB to another color space(HSV(my choice), HSL, LAB etc), and first I check it in MATLAB
I agree with Joachim that OpenCV is the most solid choice. In some cases the ITK library (https://itk.org/) can be also used, but using OpenCV is much better..
Matlab is a framework to interactive design scripts, and funtion libraries. OpenCV is a collection of functions well founded and better classified to "Image processing" and more. So posibly you can execute openCV tools from a Matlab environment. Answering your question, are not comparable. Is posible comparing Python to Matlab in some sense, because they let do similar things. And comparing functions that includes Matlab in its image processing toolbox [1] with the image processing in OpenCV [2]. To do that i recomend look those two links. And then select the execution environment, Matlab, Python or direct shell with C++. All this in the informatic low profile.
[1] https://es.mathworks.com/help/images/functionlist.html?requestedDomain=www.mathworks.com
[2] http://docs.opencv.org/3.0-beta/doc/py_tutorials/py_imgproc/py_table_of_contents_imgproc/py_table_of_contents_imgproc.html
http://docs.opencv.org/2.4/doc/tutorials/imgproc/table_of_content_imgproc/table_of_content_imgproc.html
As you know there are several (perhaps many) different ways to extract information of interest from digital color images. I come from the remote sensing community who uses satellite, airborne, and ship borne imaging to help study the earth surface (including the bottom of the ocean) and classification or feature extraction is an often main application. Like Joachim I would be interested in seeing (and if possible getting) sample image data sets and what you are trying to classify.
Pat
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Human-Computer Interaction