Watershed in Android Studio

Android/Java
1

I try use that code to segment an image, and it work in visual studio, but I need to translate to Java in Android Studio.

When I translate, I have that program:

    public Mat watershedVisual(Bitmap bmp){
    Mat img = new Mat();
    Mat rgba = new Mat();
    Utils.bitmapToMat(bmp, img);
    img.convertTo(img, CvType.CV_8UC3);
    Imgproc.cvtColor(img, rgba, Imgproc.COLOR_RGBA2RGB);
    Mat dest = new Mat();


    Mat markers = new Mat(rgba.size(),CvType.CV_8UC3, new Scalar(-1));
    markers.submat(new Rect(0,0,rgba.cols(), 5)).setTo(new Scalar(1,1,1));
    markers.submat(new Rect(0,rgba.rows()-5,rgba.cols(), 5)).setTo(new Scalar(1,1,1));
    markers.submat(new Rect(0,0,5, rgba.rows())).setTo(new Scalar(1,1,1));
    markers.submat(new Rect(rgba.cols()-5,0,5, rgba.rows())).setTo(new Scalar(1,1,1));

    int centreW = rgba.cols()/4;
    int centreH = rgba.rows()/4;

    markers.submat(new Rect((rgba.cols()/2)-(centreW/2), (rgba.rows()/2)-(centreH/2), centreW, centreH)).setTo(new Scalar(2,2,2));
    markers.convertTo(markers, Imgproc.COLOR_BGR2GRAY);



    WatershedSegmenter segmenter = new WatershedSegmenter();
    Mat wshedMask = new Mat();
    segmenter.setMarkers(markers);
    wshedMask = segmenter.process(rgba);
    Mat mask = new Mat();
    Core.convertScaleAbs(wshedMask, mask, 1, 0);
    double thresh = Imgproc.threshold(mask, mask, 1, 255, Imgproc.THRESH_BINARY);
    Core.bitwise_and(rgba, rgba, dest, mask);
    dest.convertTo(dest, CvType.CV_8UC3);
    return dest;
}

The problem is when I execute it, I have that problem:

E/cv::error(): OpenCV(3.4.11) Error: Assertion failed (src.type() == CV_8UC3 && dst.type() == CV_32SC1) in void cv::watershed(cv::InputArray, cv::InputOutputArray), file /build/3_4_pack-android/opencv/modules/imgproc/src/segmentation.cpp, line 161

I convert the image like stackoverflow, but in android there are a problem with that and I don’t know how fix it.

class WatershedSegmenter {
    public Mat markers=new Mat();

    public void setMarkers(Mat markerImage)
    {
        markerImage.convertTo(markers, CvType.CV_32SC1);
    }

    public Mat process(Mat image)
    {
        image.convertTo(image, CvType.CV_8UC3);
        markers.convertTo(markers,CvType.CV_32SC1);
        Imgproc.watershed(image,markers);
        //Log.i(TAG, "LLEGA AQUI");
        markers.convertTo(markers,CvType.CV_8UC3);
        return markers;
    }
};

Any help please?

2

related:

3

yes, but I couldn’t fix it, so I make other question with other error

4

please do you know how fix it? i don’t have idea and i need it. why if i set first image in 8uc3 and the second in 32sc1, i have an error when i executed the watershed algoritm