Java Code Examples for org.opencv.core.Core#reduce()

/**
 * Calculate the meanFeatureVector for each StudentImageCollectionEvent using the extracted featureVectors
 */
public synchronized void trainClassifier(){
    Log.i(getClass().getName(), "trainClassifier");
    // Initiate training if a StudentImageCollectionEvent has not been trained yet but all corresponding StudentImage's features have been extracted
    List<StudentImageCollectionEvent> studentImageCollectionEvents = studentImageCollectionEventDao.queryBuilder().where(StudentImageCollectionEventDao.Properties.MeanFeatureVector.isNull()).list();
    Log.i(getClass().getName(), "Count of StudentImageCollectionEvents where MeanFeatureVector is null: " + studentImageCollectionEvents.size());
    if (studentImageCollectionEvents.size() > 0){
        for (StudentImageCollectionEvent studentImageCollectionEvent : studentImageCollectionEvents){
            Long studentImagesWithoutExtractedFeatures = studentImageDao.queryBuilder()
                    .where(StudentImageDao.Properties.StudentImageCollectionEventId.eq(studentImageCollectionEvent.getId()))
                    .where(StudentImageDao.Properties.StudentImageFeatureId.eq(0))
                    .count();
            // Skip calculation of meanFeatureVector if not all features have been extracted yet
            if (studentImagesWithoutExtractedFeatures == 0){
                Mat allFeatureVectors = new Mat();
                List<StudentImage> studentImages = studentImageCollectionEvent.getStudentImages();
                for (StudentImage studentImage : studentImages){
                    List<Float> featureVectorList = gson.fromJson(studentImage.getStudentImageFeature().getFeatureVector(), new TypeToken<List<Float>>(){}.getType());
                    Mat featureVector = Converters.vector_float_to_Mat(featureVectorList);
                    allFeatureVectors.push_back(featureVector.reshape(1, 1));
                }

                Mat meanFeatureVector = new Mat();
                Core.reduce(allFeatureVectors, meanFeatureVector, 0, Core.REDUCE_AVG);
                List<Float> meanFeatureVectorList = new ArrayList<>();
                Converters.Mat_to_vector_float(meanFeatureVector.reshape(1, meanFeatureVector.cols()), meanFeatureVectorList);
                String meanFeatureVectorString = gson.toJson(meanFeatureVectorList);
                studentImageCollectionEvent.setMeanFeatureVector(meanFeatureVectorString);

                Student student = createStudent(studentImages);

                studentImageCollectionEvent.setStudent(student);
                studentImageCollectionEventDao.update(studentImageCollectionEvent);
                Log.i(getClass().getName(), "StudentImageCollectionEvent with Id " + studentImageCollectionEvent.getId() + " has been trained in classifier");
            } else {
                Log.i(getClass().getName(), "trainClassifier: Calculation of meanFeatureVector has been skipped for the StudentImageCollectionEvent: " + studentImageCollectionEvent.getId() + " studentImagesWithoutExtractedFeatures: " + studentImagesWithoutExtractedFeatures);
            }
        }
    }
}
 

public static Rect detectBorder(Mat original){
    Mat src = original.clone();
    Log.d(TAG, "1 original: " + src.toString());

    Imgproc.GaussianBlur(src, src, new Size(3, 3), 0);
    Log.d(TAG, "2.1 --> Gaussian blur done\n blur: " + src.toString());

    Imgproc.cvtColor(src, src, Imgproc.COLOR_RGBA2GRAY);
    Log.d(TAG, "2.2 --> Grayscaling done\n gray: " + src.toString());

    Mat sobelX = new Mat();
    Mat sobelY = new Mat();

    Imgproc.Sobel(src, sobelX, CvType.CV_32FC1, 2, 0, 5, 1, 0);
    Log.d(TAG, "3.1 --> Sobel done.\n X: " + sobelX.toString());
    Imgproc.Sobel(src, sobelY, CvType.CV_32FC1, 0, 2, 5, 1, 0);
    Log.d(TAG, "3.2 --> Sobel done.\n Y: " + sobelY.toString());

    Mat sum_img = new Mat();
    Core.addWeighted(sobelX, 0.5, sobelY, 0.5, 0.5, sum_img);
    //Core.add(sobelX, sobelY, sum_img);
    Log.d(TAG, "4 --> Addition done. sum: " + sum_img.toString());

    sobelX.release();
    sobelY.release();

    Mat gray = new Mat();
    Core.normalize(sum_img, gray, 0, 255, Core.NORM_MINMAX, CvType.CV_8UC1);
    Log.d(TAG, "5 --> Normalization done. gray: " + gray.toString());
    sum_img.release();

    Mat row_proj = new Mat();
    Mat col_proj = new Mat();
    Core.reduce(gray, row_proj, 1, Core.REDUCE_AVG, CvType.CV_8UC1);
    Log.d(TAG, "6.1 --> Reduce done. row: " + row_proj.toString());

    Core.reduce(gray, col_proj, 0, Core.REDUCE_AVG, CvType.CV_8UC1);
    Log.d(TAG, "6.2 --> Reduce done. col: " + col_proj.toString());
    gray.release();

    Imgproc.Sobel(row_proj, row_proj, CvType.CV_8UC1, 0, 2);
    Log.d(TAG, "7.1 --> Sobel done. row: " + row_proj.toString());

    Imgproc.Sobel(col_proj, col_proj, CvType.CV_8UC1, 2, 0);
    Log.d(TAG, "7.2 --> Sobel done. col: " + col_proj.toString());

    Rect result = new Rect();

    int half_pos = (int) (row_proj.total()/2);
    Mat row_sub = new Mat(row_proj, new Range(0, half_pos), new Range(0, 1));
    Log.d(TAG, "8.1 --> Copy sub matrix done. row: " + row_sub.toString());
    result.y = (int) Core.minMaxLoc(row_sub).maxLoc.y;
    Log.d(TAG, "8.2 --> Minmax done. Y: " + result.y);
    row_sub.release();
    Mat row_sub2 = new Mat(row_proj, new Range(half_pos, (int) row_proj.total()), new Range(0, 1));
    Log.d(TAG, "8.3 --> Copy sub matrix done. row: " + row_sub2.toString());
    result.height = (int) (Core.minMaxLoc(row_sub2).maxLoc.y + half_pos - result.y);
    Log.d(TAG, "8.4 --> Minmax done. Height: " + result.height);
    row_sub2.release();

    half_pos = (int) (col_proj.total()/2);
    Mat col_sub = new Mat(col_proj, new Range(0, 1), new Range(0, half_pos));
    Log.d(TAG, "9.1 --> Copy sub matrix done. col: " + col_sub.toString());
    result.x = (int) Core.minMaxLoc(col_sub).maxLoc.x;
    Log.d(TAG, "9.2 --> Minmax done. X: " + result.x);
    col_sub.release();
    Mat col_sub2 = new Mat(col_proj, new Range(0, 1), new Range(half_pos, (int) col_proj.total()));
    Log.d(TAG, "9.3 --> Copy sub matrix done. col: " + col_sub2.toString());
    result.width = (int) (Core.minMaxLoc(col_sub2).maxLoc.x + half_pos - result.x);
    Log.d(TAG, "9.4 --> Minmax done. Width: " + result.width);
    col_sub2.release();

    row_proj.release();
    col_proj.release();
    src.release();

    return result;
}
 

/**
 * Calculate the meanFeatureVector for each StudentImageCollectionEvent using the extracted featureVectors
 */
public synchronized void trainClassifier(){
    Log.i(getClass().getName(), "trainClassifier");
    // Initiate training if a StudentImageCollectionEvent has not been trained yet but all corresponding StudentImage's features have been extracted
    List<StudentImageCollectionEvent> studentImageCollectionEvents = studentImageCollectionEventDao.queryBuilder().where(StudentImageCollectionEventDao.Properties.MeanFeatureVector.isNull()).list();
    Log.i(getClass().getName(), "Count of StudentImageCollectionEvents where MeanFeatureVector is null: " + studentImageCollectionEvents.size());
    if (studentImageCollectionEvents.size() > 0){
        for (StudentImageCollectionEvent studentImageCollectionEvent : studentImageCollectionEvents){
            Long studentImagesWithoutExtractedFeatures = studentImageDao.queryBuilder()
                    .where(StudentImageDao.Properties.StudentImageCollectionEventId.eq(studentImageCollectionEvent.getId()))
                    .where(StudentImageDao.Properties.StudentImageFeatureId.eq(0))
                    .count();
            // Skip calculation of meanFeatureVector if not all features have been extracted yet
            if (studentImagesWithoutExtractedFeatures == 0){
                Mat allFeatureVectors = new Mat();
                List<StudentImage> studentImages = studentImageCollectionEvent.getStudentImages();
                for (StudentImage studentImage : studentImages){
                    List<Float> featureVectorList = gson.fromJson(studentImage.getStudentImageFeature().getFeatureVector(), new TypeToken<List<Float>>(){}.getType());
                    Mat featureVector = Converters.vector_float_to_Mat(featureVectorList);
                    allFeatureVectors.push_back(featureVector.reshape(1, 1));
                }

                Mat meanFeatureVector = new Mat();
                Core.reduce(allFeatureVectors, meanFeatureVector, 0, Core.REDUCE_AVG);
                List<Float> meanFeatureVectorList = new ArrayList<>();
                Converters.Mat_to_vector_float(meanFeatureVector.reshape(1, meanFeatureVector.cols()), meanFeatureVectorList);
                String meanFeatureVectorString = gson.toJson(meanFeatureVectorList);
                studentImageCollectionEvent.setMeanFeatureVector(meanFeatureVectorString);

                Student student = createStudent(studentImages);

                studentImageCollectionEvent.setStudent(student);
                studentImageCollectionEventDao.update(studentImageCollectionEvent);
                Log.i(getClass().getName(), "StudentImageCollectionEvent with Id " + studentImageCollectionEvent.getId() + " has been trained in classifier");
            } else {
                Log.i(getClass().getName(), "trainClassifier: Calculation of meanFeatureVector has been skipped for the StudentImageCollectionEvent: " + studentImageCollectionEvent.getId() + " studentImagesWithoutExtractedFeatures: " + studentImagesWithoutExtractedFeatures);
            }
        }
    }
}
 

private void computePsi(){
    Core.reduce(Gamma, Psi, 0, Core.REDUCE_AVG);
}