Question

I am using OpenCV for some of my image processing related codes. I am stuck at a particular point.

In a particular code, I need to compare the pixel values of an image based on it`s CIE L*a*b* values. I know for BGR we can do something like this :

// img is the image, suppose
uchar *data = ( uchar* )img->imageData;

for(i=0; i< img->height; i++)
    for(j=0; j<img->width; j++)
     {
         b = data[i*img->widthStep + j*img->nChannels + 0];
         g = data[i*img->widthStep + j*img->nChannels + 1];
         r = data[i*img->widthStep + j*img->nChannels + 2]; 
     }

My question is, can we do something similar for l*a*b* color model also ? I mean at first I convert the image by

cvCvtColor(img, lab, CV_BGR2Lab);

then accessing the individual channel info like in the above case (since L*a*b* is also made up of 3 channels, if am not wrong), can we get the pixel values ??

The reason I am asking this is I tried to implement it, but I`m getting some real weird values like very high negative numbers ( I used int datatype instead of uchar since I was unsure about the range ‘a’ and ‘b’ can have ) whereas in Wikipedia I saw that ‘L’ can have values only in the range of 0 to 100. So how exactly can I get those pixel values ??

Thanks in advance !!

Answer 1

Surely, you can do that, why not? Be aware however, that output is somewhat specific (since a and b can be negative):

On output 0≤L≤100, -127≤a≤127, -127≤b≤127
The values are then converted to the destination data type:
    8-bit images:
        L <- L*255/100, a <- a + 128, b <- b + 128

note that L is also scaled to 0-255 range (instead of 0-100). See more here.

BTW how can you get negative numbers if your data type is unsigned char?