I am doing work in image retrieval using brain image (Gray scale image). My concept is Texture, color and shapes feature extraction process going to do.
Is it possible for finding color feature in brain image but it is gray scale image not color image.
Please suggest me friends or any idea friends...
see this paper
https://www.researchgate.net/publication/260714141_Rigid_Point_Feature_Registration_for_MRI_Brain_Images_based_on_Fuzzy_Transform
Article Rigid Point Feature Registration for MRI Brain Images based ...
Thanks for your information Athiraja Atheeswaran sir and Mohammed y kamil sir...
Mr. Pandian, I dont think you can extract color information from a gray level image. For example, in MATLAB, image is represented as I(m,n,p), where m,n comes from resolution of image and p represents the number of channel the image has.In opencv the image representations differs. Suppose, you are using RGB image means p would be 3, if you transformed the RGB to HSV/LAB/HSI, p would be 3, if you are transforming it to CMYK means p would be 4, thus the number of channels may or may not vary but surely the content of pixels i.e. pixel intensity varies.
I will show you an example MATLAB code,
I=imread('file.jpeg'); % Reading a RGB file
R=I(:,:,1); % Extracting Red channel Image
G=I(:,:,2); % Extracting Green channel Image
B=I(:,:,3);% Extracting Blue channel Image
if you use imshow for variables R,G,B it may look like gray scale image but it is not.
similary if you use, IG=rgb2gray(I); and viewed it using imshow it may look like R/G/B images but it is gray level image.
If a 2D Image Array does not represent color means how it is possible to extract color information from it.
Thus extracting Texture, shapes, edges from gray image is possible but color information is not possible since finding which color channle contibuted to gray level intensity is tough and impossible as far as i know
I'm not sure exactly what you mean by color feature extraction, could you maybe provide an example?
My initial response is to say no. As others pointed out, there is no color information present other than the grayscale values, so there's no way of recovering color in any colorspace (RGB,HSI,Lab,etc.). However, you could extract features based on the grayscale values directly, such as a histogram of the grayscale values and various measurements calculated from the histogram (mean, variance, entropy, kurtosis, etc.).
shall i explain in person? I think that there are possibilities in your doubts
You can not extract color features using gray level image but you can extract texture features from gray image. after you transformed RGB to Gray in MATLAB software you can acquire GLCM and GLRM matrix and then extract texture features with this matrix.
Color Features are those statistical features calculated based on the histogram (for example).
For color images (True images) , it could be extracted from all channels while in gray it has to be extracted only from the gray image.
You can extract color feature from gray scale image also using quantization method.
In color image you have 3channels but gray scale image 1 channel. Even though intensities levels is varies. color image has 65536 intensity level (display color) because 16 bit image, gray scale image has 256 intensity level (display color) because 8 bit image. So no need to use all color for feature extraction. You can extract only required color for from any image using quantization. Further reference read following papers:
A novel method for image retrieval based on structure elements’ descriptor
Rotation and scale invariant hybrid image descriptor and
retrieval
Note : If you have any more doubt message me.
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing