How to apply CLAHE on RGB color images

Question

CLAHE is Contrast Limited Adaptive Histogram Equalization and a source in C can be found at http://tog.acm.org/resources/GraphicsGems/gemsiv/clahe.c

So far I have only seen some examples/tutorials about applying CLAHE on grayscale images so is it possible to apply CLAHE on color images (such as RGB 3 channles images)? If yes, how?

Kulturkampf

not possible directly. maybe convert to LAB or HSV, apply clahe on L and convert back. and you can use it from opencv, too. – Yttria

@Yttria thanks and your comment could be an answer – Kulturkampf

possible duplicate of simple illumination correction in images openCV c++ – Wun

^^ yes, B. that's where i got the idea. – Yttria

This link will help you #24341614 – Markel

Answer 1

Conversion of RGB to LAB(L for lightness and a and b for the color opponents green–red and blue–yellow) will do the work. Apply CLAHE to the converted image in LAB format to only Lightness component and convert back the image to RGB. Here is the snippet.

bgr = cv2.imread(image_path)

lab = cv2.cvtColor(bgr, cv2.COLOR_BGR2LAB)

lab_planes = cv2.split(lab)

clahe = cv2.createCLAHE(clipLimit=2.0,tileGridSize=(gridsize,gridsize))

lab_planes[0] = clahe.apply(lab_planes[0])

lab = cv2.merge(lab_planes)

bgr = cv2.cvtColor(lab, cv2.COLOR_LAB2BGR)

bgr is the final RGB image obtained after applying CLAHE.

Quasi

cv2.split isn't required as OpenCV in Python uses NumPy arrays. Once you create the CLAHE object, just do lab[...,0] = clahe.apply(lab[...,0]). You can also remove cv2.merge – Melodrama

This method sometimes gives bad results. See point 4 of #56906092. Would you have any idea @NikhilVM? – Passementerie

@Passementerie One has to try varying the parameters clipLimit and tileGridSize. In the question linked above, the OP has used grid size of 100. – Gotten

Answer 2

img = cv2.imread(img_path, cv2.IMREAD_COLOR)
img = cv2.cvtColor(img, cv2.COLOR_RGB2Lab)

#configure CLAHE
clahe = cv2.createCLAHE(clipLimit=10,tileGridSize=(8,8))

#0 to 'L' channel, 1 to 'a' channel, and 2 to 'b' channel
img[:,:,0] = clahe.apply(img[:,:,0])

img = cv2.cvtColor(img, cv2.COLOR_Lab2RGB)

cv2.imshow("image", img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Macfadyn

Answer 3

this for c#, apply clahe for rgb images.

    private static Image<Bgr, byte> appy_CLAHE( Image<Bgr, byte> imgIn , int clipLimit=2, int tileSize=25)
    {
        var imglabcol = new Image<Lab, byte>(imgIn.Size);
        var imgoutL = new Image<Gray, byte>(imgIn.Size);  

        var imgoutBGR = new Image<Bgr, byte>(imgIn.Size);  

        //clahe filter must be applied on luminance channel or grayscale image
        CvInvoke.CvtColor(imgIn, imglabcol, ColorConversion.Bgr2Lab, 0); 

        CvInvoke.CLAHE(imglabcol[0], clipLimit, new Size(tileSize, tileSize), imgoutL);
        imglabcol[0] = imgoutL; // write clahe results on Lum channel into image

        CvInvoke.CvtColor(imglabcol, imgoutBGR, ColorConversion.Lab2Bgr, 0);

        return imgoutBGR;
    }

Decoration