Solace's profile - activity

If I use addweighted method to overlay the image in the live camera preview, will the overlayed image also be added automatically to the photo I capture?

@berak So if I wanted to draw it separately on the captured image, I wouldn't know the exact and precise locations of points and may be sizeas well.

@berak I want the polygon to be at the exact same place in the captured image as it was in the live camera preview. For example, say the overlayed image was the boundary of a closed shape like a human's face (like in this image.), and in the live camera preview it is used to to guide a user where to place their face in the camera frame. Now in the image captured, I would want to merge that boundary lines into the captured image, and I would want them to be exactly where they were in the camera preview; I would not want the captured image having the face on one position, and the overlayed face-boundary displaced to somewhere else.

The transparent image I want to overlay is just the boundary of a polygon in black color, the rest of the image is transparent. Like this:

Using OpenCV4Android, I want to overlay it over live camera preview, such that it is overlayed on the image captured as well (i.e. the polygon is blended over the image captured, I mean the polygon should become a part of the image captured.)

How can I achieve that?

ADDENDUM:

I would simply add the polygon to the live camera preview by overlaying the JavaCameraView (which is the widget which displays live camera preview/frames) with an ImageView (Android's image container widget) containing the transparent image to overlay; by placing them both sequentially in a FrameLayout, which automatically Z-orders widgets added to it. Then I would separately add the image to overlay into to captured photo.

But the challenge here is:

The polygon in the live camera preview/frames is a guide. Just like the following image has a guide for positioning the face before taking the picture.

Then when the photo is captured, I want the polygon to be overlayed at EXACTLY the position on which it was when the photo was captured.

So how do I achieve this?

@sturkmen Sorry I was wrong in previous comment. nimages parameter of histPrepareImages() is not what I passed to calcHist() as numberOfBins. Rather, after carefully comparing the parameters of histPrepareImages() here, calcHist() here for C++, and the ones shown in Eclipse (IDE) Javadoc, I learnt that Android API does not have a int nimages argument for calcHist(). For Android API, the calcHist() is Imgproc.calcHist(images, channels, mask, hist, histSize, ranges);

@sturkmen That code is just like mine here in this question. But looking at the definition of histPrepareImages() - the method in which I have got the error of j < nimages according the the OpenCV Error statement in logcat - given on line 115 of this class tells me that nimages is the second arg, which is numberOfBins in my code. So it means there is something with my numOfBins argument to calcHist(). Can you look into Line 115 of that file and see what might have caused j to be less than numOfBins I passed and any hint on what is going wrong? I don't understand C++ or OpenCV well.

@sturkmen Thank you. I tried to write an Android version from the tutorial you pointed me too. But I am getting an here. If you have some time, could you have a look into my new question?

I am going to do some web search to understand those steps a bit better. If I still don't get it, I will post more specific questions. Thank you so much so far.

@sturkmen I am not able to understand step 5 in their explanation. How do they know that they have to use 512 and 400 as width and height values? Actually all steps after this one is confusing.

I have written code to read in an image from gallery, then apply inRange() on it to find blue colored blob and create mask, then used findContours() to find contours, and the drawContours() to draw the contours, then boundingRect() to find a rectangle around the first contour, and then submat() to extract out the submatrix. Then I used Mat.zeroes() to create a mask from that rectangular submat. Then I applied Imgproc.calcHist() to create a histogram, and passed in the Mask as the 3rd argument so that the histogram calculated is that of the area inside the contour only.

Following is a screenshot from my Genymotion emulator, showing the original Mat with green contours around blue detected blobs and then the submatrix I mentioned and then the histogram created by calcHist(), which is black vertical line The question is why is it like this, where is my histogram graph?

Code:

//Get the image from file path
            rgbaMat = new Mat();
            rgbaMatCopy = new Mat();
            rgbaMat = Highgui.imread(filePath);
            rgbaMatCopy = Highgui.imread(filePath);

            //Preprocessing - color space conversion
            Imgproc.cvtColor(rgbaMat, rgbaMat, Imgproc.COLOR_BGR2RGB);
            Imgproc.cvtColor(rgbaMatCopy, rgbaMatCopy, Imgproc.COLOR_BGR2RGB);
            Mat hsvMat = new Mat();
            Imgproc.cvtColor(rgbaMatCopy, hsvMat, Imgproc.COLOR_RGB2HSV);

            //Make the mask from the detected blue colored blob(s)
            Scalar lowerThreshold = new Scalar(100, 146, 148);
            Scalar upperThreshold = new Scalar(134, 255, 255);
            Mat mask = new Mat();
            Core.inRange(hsvMat, lowerThreshold, upperThreshold, mask);

            //Preprocessing
            Mat dilatedMask = new Mat();
            Imgproc.dilate(mask, dilatedMask, new Mat());
            Imgproc.dilate(dilatedMask, dilatedMask, new Mat());

            //Find the contours
            contours = new ArrayList<>();
            Imgproc.findContours(dilatedMask, contours, new Mat(), Imgproc.RETR_EXTERNAL, Imgproc.CHAIN_APPROX_SIMPLE);

            //Draw the contours
            for (int contourIdx = 0; contourIdx < contours.size(); contourIdx++) {
                Imgproc.drawContours(rgbaMatCopy, contours, contourIdx, new Scalar(0, 255, 0), 1);
            }

            //Convert the mat with contours drawn to bitmap and display in imageView
            Bitmap theBitmap = Bitmap.createBitmap(rgbaMatCopy.cols(), rgbaMatCopy.rows(),
                    Bitmap.Config.ARGB_8888);
            Utils.matToBitmap(rgbaMatCopy, theBitmap);
            imageView.setImageBitmap(theBitmap);

            //Find and cut out the smallest bounding rectangle around the first contour from the list of contours
            MatOfPoint firstContour = contours.get(0);
            Rect firstContourBoundingRect = Imgproc.boundingRect(firstContour);
            Mat objectSubmat = new Mat();
            objectSubmat = rgbaMat.submat(firstContourBoundingRect);

            Bitmap objectSubmatBitmap = Bitmap.createBitmap(objectSubmat.cols(), objectSubmat.rows(), Bitmap.Config.ARGB_8888);
            Utils.matToBitmap(objectSubmat, objectSubmatBitmap);
            imageViewOne.setImageBitmap(objectSubmatBitmap);

            //Create mask of the region inside the contour
            Mat regionMask = Mat.zeros(objectSubmat.rows(), objectSubmat.cols(), CvType.CV_8UC1);
            List<MatOfPoint> firstList = new ArrayList<MatOfPoint>();
            firstList.add(firstContour);
            Imgproc.drawContours(regionMask, firstList, 0, new Scalar(255), -1);

            //Preprocessing - color space conversion
            Mat regionHsv = new Mat();
            Imgproc.cvtColor(objectSubmat, regionHsv, Imgproc.COLOR_RGB2HSV);


            //Calculate Histogram
            List<Mat> sourceImageList = new ArrayList<Mat>();
            sourceImageList.add(regionHsv);
            Mat histogram = new Mat();
            //histogramMat.convertTo(histogramMat, CvType.CV_8UC2);
            Imgproc.calcHist(sourceImageList, new MatOfInt(2), regionMask, histogram, new MatOfInt(256), 
                    new MatOfFloat(0f, 256f));


        Highgui.imwrite("/mnt/sdcard/histogram.bmp", histogram);// check
        Mat hish = Highgui.imread("/mnt/sdcard/histogram.bmp");
        Bitmap histogramBitmap = 
                Bitmap.createBitmap(hish.cols(), hish.rows(), Bitmap.Config.ARGB_8888);
        Utils.matToBitmap(hish, histogramBitmap);
        imageViewTwo.setImageBitmap(histogramBitmap);

I am trying to calculate the histogram from this image and trying to convert it to a Bitmap and display it.

And I am getting the following error. The question is why, and how can I fix it?

06-26 14:07:18.972: E/cv::error()(1278): OpenCV Error: Assertion failed (src.type() == CV_8UC1 || src.type() == CV_8UC3 || src.type() == CV_8UC4) in void Java_org_opencv_android_Utils_nMatToBitmap2(JNIEnv *, jclass, jlong, jobject, jboolean), file /hdd2/buildbot/slaves/slave_ardbeg1/50-SDK/opencv/modules/java/generator/src/cpp/utils.cpp, line 98
06-26 14:07:18.992: A/libc(1278): Fatal signal 11 (SIGSEGV) at 0x997fcff4 (code=2), thread 1278 (tests.testfive)

Following is the relevant part of my code.

    //Get the image from file path
    rgbaMat = new Mat();
    rgbaMat = Highgui.imread(filePath);

    Bitmap objectSubmatBitmap = Bitmap.createBitmap(rgbaMat.cols(), rgbaMat.rows(), Bitmap.Config.ARGB_8888);
    Utils.matToBitmap(rgbaMat, objectSubmatBitmap);
    imageViewOne.setImageBitmap(objectSubmatBitmap);

    //Preprocessing - color space conversion
    Imgproc.cvtColor(rgbaMat, rgbaMat, Imgproc.COLOR_BGR2RGB);
    Mat hsvMat = new Mat();
    Imgproc.cvtColor(rgbaMat, hsvMat, Imgproc.COLOR_RGB2HSV);


    //Calculate Histogram
    List<Mat> sourceImageList = new ArrayList<Mat>();
    sourceImageList.add(hsvMat);
    Mat histogram = new Mat();
    //histogramMat.convertTo(histogramMat, CvType.CV_8UC1);
    Imgproc.calcHist(sourceImageList, new MatOfInt(2), new Mat(), histogram, new MatOfInt(256), 
                    new MatOfFloat(0f, 256f));

    //Highgui.imwrite("/mnt/sdcard/histogram.bmp", histogram);// check
    Bitmap histogramBitmap = 
    Bitmap.createBitmap(histogram.cols(), histogram.rows(), Bitmap.Config.ARGB_8888);
    Utils.matToBitmap(histogram, histogramBitmap);
    imageViewOne.setImageBitmap(histogramBitmap);

@sturkmen Thank you. I posted the complete code and the original image in the question.

I started with this image:

[![enter image description here][1]]

Then I applied Canny Edge Detector like:

Mat originalMatGreyScale = new Mat();
Imgproc.cvtColor(originalPhotoMat, originalMatGreyScale, Imgproc.COLOR_BGR2GRAY);
Mat edgesMat = new Mat();
Imgproc.Canny(originalMatGreyScale, edgesMat , 50, 70);

I got:

[![enter image description here][2]]

Then I found a list of contours (contours) by Imgproc.findContours().Then I did some coding to (1)find the area of largest contour (maximumContourArea) (2) Remove from contours any contour which has an area less than the maximumContourArea. Code given as follows.

What I was expecting was that all those little smudges and noise should be removed by this, but instead I still got this:

[![enter image description here][3]]

Moreover, the Log.i() messages in the following code print this:

Number of contours initially: 27

Number of contours after processing: 27

List<MatOfPoint> contours = new ArrayList<MatOfPoint>();
Imgproc.findContours(edgesMap, contours, new Mat(), Imgproc.RETR_EXTERNAL, Imgproc.CHAIN_APPROX_SIMPLE);


Log.i(TAG, "Number of contours initially: " + contours.size());//check

double maximumContourArea = 0;

Iterator<MatOfPoint> contoursIterator = contours.iterator();

while(contoursIterator.hasNext()) {
    MatOfPoint nextContour = contoursIterator.next();
    double nextContourArea = Imgproc.contourArea(nextContour);

    if (nextContourArea > maximumContourArea) {
        maximumContourArea = nextContourArea;
    }
}

while(contoursIterator.hasNext()) {
    MatOfPoint nextContour = contoursIterator.next();
    if (Imgproc.contourArea(nextContour) < maximumContourArea*(10 / 100)) {
        contours.remove(contours.indexOf(nextContour));
    }
}

Log.i(TAG, "Number of contours after processing: " + contours.size());//check

This helped get rid of the garbage in the mask, but still not working correctly. Tried a couple of things, which have confused me more. Can you see the edit in the question?

I started by reading in a Mat.

Then I converted it to Greyscale and applied Imgproc.canny() to it, getting the following mask.

Then I used Imgproc.findContours() to find the contours, Imgproc.drawContours(), and Core.putText() to label the contours with numbers

Then I did Rect boundingRect = Imgproc.boundingRect(contours.get(0)); Mat submatrix = new Mat(); submatrix = originalMat.submat(boundingRect); to get following submatrix:

So far so good. The Problem starts hereafter:

NOW I NEEDED A MASK OF THE submatrix. So I decided to use Imgproc.drawContours() to get the mask:

Mat mask = new Mat(submatrix.rows(), submatrix.cols(), CvType.CV_8UC1);
        List<MatOfPoint> contourList = new ArrayList<>();
        contourList.add(contours.get(0));
        Imgproc.drawContours(mask, contourList, 0, new Scalar(255), -1);

I got the following mask:

WHAT I WAS EXPECTING was a filled (in white color) diamond shape on black background.

WHy am I getting this unexpected result?

Given an image Mat, and a contour (which is a MatOfPoint) in it, how can I create a ROI (Region Of Interest)/submat?

I can see three interesting methods on docs of Mat,

Mat submat(int rowStart, int rowEnd, int colStart, int colEnd) Extracts a rectangular submatrix.

Mat submat(Range rowRange, Range colRange) Extracts a rectangular submatrix.

Mat submat(Rect roi) Extracts a rectangular submatrix.

• Is there a way to find out rowStart, rowEnd, colStart and colEnd from the contour?

or

• Is there a way to get rowRange and colRange from the contour?

or

• Can I make a Rect from the contour?