Using custom color palette with kmeans color clustering algorithm

Android/Java
,
1 
         bitmap = MediaStore.Images.Media.getBitmap(this.getContentResolver(), 
          data.getData());
            ny = new Mat();
            Utils.bitmapToMat(bitmap, ny);

       Mat bgrImage = new Mat();
        Imgproc.cvtColor(ny, bgrImage, Imgproc.COLOR_RGB2BGR);
        bgrImage.convertTo(bgrImage, CvType.CV_32F, 1.0 / 255.0);


        // Convert the image to grayscale canny ımage
         Mat gres = new Mat();
         Imgproc.cvtColor(ny, gres, Imgproc.COLOR_BGR2GRAY);


        // Compute median intensity
        double medianIntensity = Core.mean(gres).val[0];


        // Compute initial lower and upper thresholds
         double lowerThreshold = 0.8 * medianIntensity;
         double upperThreshold = 1.6 * medianIntensity;

        // Apply Canny edge detection with the computed thresholds
         Mat edges = new Mat();
         Imgproc.Canny(gres, edges, lowerThreshold, upperThreshold);

        // Find contours
        List<MatOfPoint> contours = new ArrayList<>();
        Mat hierarchy = new Mat();
        Imgproc.findContours(edges, contours, hierarchy, 
        Imgproc.THRESH_BINARY, Imgproc.CHAIN_APPROX_SIMPLE);

        // Reshape the image for k-means clustering
        Mat reshapedImage = bgrImage.reshape(1, bgrImage.rows() * 
         bgrImage.cols());

        // Perform k-means clustering
        Mat labels = new Mat();
        int K = 10;
        Mat centers = new Mat(K, reshapedImage.cols(), reshapedImage.type());

        TermCriteria criteria = new 
           TermCriteria(TermCriteria.EPTermCriteria.MAX_ITER, 10, 2.0);
        Core.kmeans(reshapedImage, K, labels, criteria, 1, 
         Core.KMEANS_PP_CENTERS, centers);

        // Reconstruct the clustered image
        Mat rec = new Mat(bgrImage.rows() * bgrImage.cols(), 3, CvType.CV_32F);
        for (int i = 0; i < reshapedImage.rows(); i++) {
            int j = (int) labels.get(i, 0)[0];
            centers.row(j).copyTo(rec.row(i));
        }

        // Reshape the reconstructed image to original size/type
        rec = rec.reshape(3, bgrImage.rows());
        rec.convertTo(rec, CvType.CV_8U, 255);
        Imgproc.cvtColor(rec, rec, Imgproc.COLOR_BGR2RGB);

        Mat biletra = new Mat();
        Imgproc.bilateralFilter(rec, biletra, 50, 50, 50, Core.BORDER_DEFAULT);
        Imgproc.drawContours(biletra,contours,-1 ,new Scalar (1,1,1),1);

        // Convert the processed image to Bitmap and display it
        Bitmap resultBitmap = Bitmap.createBitmap(rec.cols(), rec.rows(), 
        Bitmap.Config.ARGB_8888);
        Utils.matToBitmap(biletra, resultBitmap);
        anaekran.setImageBitmap(resultBitmap);

this is the code ı am using for color quantization it works but ı want to add custom color palettes and apply quantizaiton to the colors according to custom palettes is it possible with opencv

2

just saying, using kmeans for quantization comes with some drawbacks:

  • it’s local. the centers depend on the image content. a different image will give different centers
  • the centers (or labels) are in random order. you might have label[2] == red in one image and green in another

but ı want to add custom color palettes

have a look at color maps in opencv and lookup tables

3

so basiclly ı have to create custom colormaps and then ı have to apply colormap to the final output if didnt get this wrong

4

there are no real palette images in opencv
(this is a computer-vision library)

btw, you never mentioned, what you want to achieve here

5

Pallete Shader by GreenF0x actually this is what i am trying to achieve as you can see you can rearrenge the colors of a image with a color palette and i assumed this application achieves this by applying custom palettes to kmeans algorithm which seems like not possible