public void process(BufferedImage image) {
setInputImage(image);
T gray = ConvertBufferedImage.convertFromSingle(image, null, imageType);
ss.setImage(gray);
gui.reset();
for (int i = 0; i < ss.getTotalScales(); i++) {
ss.setActiveScale(i);
double scale = ss.getCurrentScale();
T scaledImage = ss.getScaledImage();
BufferedImage b = ConvertBufferedImage.convertTo(scaledImage, null);
gui.addImage(b, String.format("Scale %6.2f", scale));
}
processedImage = true;
}
private synchronized void setLevel(int level) {
// System.out.println("level "+level);
if (level > 0) {
ImageSingleBand small = ss.getLayer(level - 1);
ImageSingleBand enlarge =
GeneralizedImageOps.createSingleBand(
small.getClass(), ss.getInputWidth(), ss.getInputHeight());
DistortImageOps.scale(small, enlarge, TypeInterpolate.NEAREST_NEIGHBOR);
// if the size isn't the same null it so a new image will be declared
if (levelImage != null
&& (levelImage.getWidth() != enlarge.width || levelImage.getHeight() != enlarge.height)) {
levelImage = null;
}
levelImage = ConvertBufferedImage.convertTo(enlarge, levelImage);
double scale = ss.getScale(level - 1);
levelPoints.clear();
for (ScalePoint p : points) {
if (p.scale == scale) {
levelPoints.add(p);
}
}
} else {
levelPoints.clear();
levelPoints.addAll(points);
}
this.activeLevel = level;
}
/**
* ****************************************************************************************************************************************
*/
private static List<List<PointIndex_I32>> getCandidates(
BufferedImage image,
int blurRadius,
float threshLow,
float threshHigh,
double toleranceDist,
double toleranceAngle,
boolean dynamicThreshold) {
List<List<PointIndex_I32>> candidates = new ArrayList<List<PointIndex_I32>>();
ImageFloat32 input = ConvertBufferedImage.convertFromSingle(image, null, ImageFloat32.class);
ImageUInt8 binary = new ImageUInt8(input.width, input.height);
// Finds edges inside the image
CannyEdge<ImageFloat32, ImageFloat32> canny =
FactoryEdgeDetectors.canny(
blurRadius, false, dynamicThreshold, ImageFloat32.class, ImageFloat32.class);
canny.process(input, threshLow, threshHigh, binary);
List<Contour> contours = BinaryImageOps.contour(binary, rule, null);
for (Contour c : contours) {
// Only the external contours are relevant.
List<PointIndex_I32> vertices =
ShapeFittingOps.fitPolygon(c.external, true, toleranceDist, toleranceAngle, 100);
candidates.add(vertices);
}
return candidates;
}
/**
* Detects lines inside the image using different types of Hough detectors
*
* @param image Input image.
* @param imageType Type of image processed by line detector.
* @param derivType Type of image derivative.
*/
public static <T extends ImageSingleBand, D extends ImageSingleBand> void detectLines(
BufferedImage image, Class<T> imageType, Class<D> derivType) {
// convert the line into a single band image
T input = ConvertBufferedImage.convertFromSingle(image, null, imageType);
// Comment/uncomment to try a different type of line detector
DetectLineHoughPolar<T, D> detector =
FactoryDetectLineAlgs.houghPolar(
new ConfigHoughPolar(3, 30, 2, Math.PI / 180, edgeThreshold, maxLines),
imageType,
derivType);
// DetectLineHoughFoot<T,D> detector = FactoryDetectLineAlgs.houghFoot(
// new ConfigHoughFoot(3, 8, 5, edgeThreshold,maxLines), imageType, derivType);
// DetectLineHoughFootSubimage<T,D> detector = FactoryDetectLineAlgs.houghFootSub(
// new ConfigHoughFootSubimage(3, 8, 5, edgeThreshold,maxLines, 2, 2), imageType, derivType);
List<LineParametric2D_F32> found = detector.detect(input);
// display the results
ImageLinePanel gui = new ImageLinePanel();
gui.setBackground(image);
gui.setLines(found);
gui.setPreferredSize(new Dimension(image.getWidth(), image.getHeight()));
listPanel.addItem(gui, "Found Lines");
}
public static BufferedImage standard(ImageSingleBand<?> src, BufferedImage dst) {
if (src.getDataType().isInteger()) {
ImageInteger srcInt = (ImageInteger) src;
if (src.getDataType().isSigned()) {
double max = GImageStatistics.maxAbs(srcInt);
return colorizeSign(srcInt, dst, (int) max);
} else {
if (src.getDataType().getNumBits() == 8) {
dst = ConvertBufferedImage.convertTo((ImageUInt8) src, dst);
} else {
double max = GImageStatistics.maxAbs(srcInt);
dst = grayUnsigned(srcInt, dst, (int) max);
}
}
} else if (ImageFloat32.class.isAssignableFrom(src.getClass())) {
ImageFloat32 img = (ImageFloat32) src;
float max = ImageStatistics.maxAbs(img);
boolean hasNegative = false;
for (int i = 0; i < img.getHeight(); i++) {
for (int j = 0; j < img.getWidth(); j++) {
if (img.get(j, i) < 0) {
hasNegative = true;
break;
}
}
}
if (hasNegative) return colorizeSign(img, dst, (int) max);
else return grayMagnitude((ImageFloat32) src, dst, max);
}
return dst;
}
public void process(final BufferedImage buffLeft, final BufferedImage buffRight) {
imageLeft.reshape(buffLeft.getWidth(), buffLeft.getHeight());
imageRight.reshape(buffRight.getWidth(), buffRight.getHeight());
grayLeft.reshape(buffLeft.getWidth(), buffLeft.getHeight());
grayRight.reshape(buffRight.getWidth(), buffRight.getHeight());
ConvertBufferedImage.convertFromMulti(buffLeft, imageLeft, true, imageType);
ConvertBufferedImage.convertFromMulti(buffRight, imageRight, true, imageType);
SwingUtilities.invokeLater(
new Runnable() {
public void run() {
panel.setImages(buffLeft, buffRight);
processedImage = true;
doRefreshAll();
}
});
}
public static void main(String args[]) {
// load and convert the image into a usable format
BufferedImage image = UtilImageIO.loadImage("../data/applet/shapes02.png");
ImageFloat32 input = ConvertBufferedImage.convertFromSingle(image, null, ImageFloat32.class);
ShowImages.showWindow(image, "Original");
fitCannyEdges(input);
fitCannyBinary(input);
fitBinaryImage(input);
}
public void process(final BufferedImage input) {
setInputImage(input);
this.input = input;
workImage = ConvertBufferedImage.convertFromSingle(input, null, imageType);
scaledIntensity = new ImageFloat32(workImage.width, workImage.height);
pyramid.setImage(workImage);
SwingUtilities.invokeLater(
new Runnable() {
public void run() {
setPreferredSize(new Dimension(input.getWidth(), input.getHeight()));
processedImage = true;
}
});
doRefreshAll();
}
public static void main(String args[]) {
BufferedImage input = UtilImageIO.loadImage("../data/evaluation/sunflowers.png");
// BufferedImage input = UtilImageIO.loadImage("../data/evaluation/shapes01.png");
ImageFloat32 gray = ConvertBufferedImage.convertFromSingle(input, null, ImageFloat32.class);
SiftDetector alg = FactoryInterestPointAlgs.siftDetector(new ConfigSiftDetector(3, 10, 150, 5));
SiftImageScaleSpace imageSS = new SiftImageScaleSpace(1.6f, 5, 4, false);
imageSS.constructPyramid(gray);
imageSS.computeFeatureIntensity();
alg.process(imageSS);
System.out.println("total features found: " + alg.getFoundPoints().size());
VisualizeFeatures.drawScalePoints(
input.createGraphics(),
alg.getFoundPoints().toList(),
BoofDefaults.SCALE_SPACE_CANONICAL_RADIUS);
ListDisplayPanel dog = new ListDisplayPanel();
for (int i = 0; i < alg.ss.dog.length; i++) {
int scale = i % (alg.ss.numScales - 1);
int octave = i / (alg.ss.numScales - 1);
BufferedImage img = VisualizeImageData.colorizeSign(alg.ss.dog[i], null, -1);
dog.addImage(img, octave + " " + scale);
}
ListDisplayPanel ss = new ListDisplayPanel();
for (int i = 0; i < alg.ss.scale.length; i++) {
int scale = i % alg.ss.numScales;
int octave = i / alg.ss.numScales;
BufferedImage img = VisualizeImageData.grayMagnitude(alg.ss.scale[i], null, 255);
ss.addImage(img, octave + " " + scale);
}
ShowImages.showWindow(dog, "Octave DOG");
ShowImages.showWindow(ss, "Octave Scales");
ShowImages.showWindow(input, "Found Features");
System.out.println("Done");
}
/**
* Detects segments inside the image
*
* @param image Input image.
* @param imageType Type of image processed by line detector.
* @param derivType Type of image derivative.
*/
public static <T extends ImageSingleBand, D extends ImageSingleBand> void detectLineSegments(
BufferedImage image, Class<T> imageType, Class<D> derivType) {
// convert the line into a single band image
T input = ConvertBufferedImage.convertFromSingle(image, null, imageType);
// Comment/uncomment to try a different type of line detector
DetectLineSegmentsGridRansac<T, D> detector =
FactoryDetectLineAlgs.lineRansac(40, 30, 2.36, true, imageType, derivType);
List<LineSegment2D_F32> found = detector.detect(input);
// display the results
ImageLinePanel gui = new ImageLinePanel();
gui.setBackground(image);
gui.setLineSegments(found);
gui.setPreferredSize(new Dimension(image.getWidth(), image.getHeight()));
listPanel.addItem(gui, "Found Line Segments");
}
/**
* Renders a colored image where the color indicates the sign and intensity its magnitude. The
* input is divided by normalize to render it in the appropriate scale.
*
* @param src Input single band image.
* @param dst Where the image is rendered into. If null a new BufferedImage will be created and
* return.
* @param normalize Used to normalize the input image. If <= 0 then the max value will be used
* @return Rendered image.
*/
public static BufferedImage colorizeSign(
ImageSingleBand src, BufferedImage dst, double normalize) {
dst = checkInputs(src, dst);
if (normalize <= 0) {
normalize = GImageStatistics.maxAbs(src);
}
if (normalize == 0) {
// sets the output to black
ConvertBufferedImage.convertTo(src, dst, true);
return dst;
}
if (src.getClass().isAssignableFrom(ImageFloat32.class)) {
return colorizeSign((ImageFloat32) src, dst, (float) normalize);
} else {
return colorizeSign((ImageInteger) src, dst, (int) normalize);
}
}
public static void main(String[] args) throws IOException {
if (args.length == 0) {
System.out.println(
"usage: crop-objects [OPTION]... FILE [DIR]\n"
+ "crops detected objects from image FILE and writes their subimages to files. \n"
+ "Can specify the DIR in which to create the files, otherwise subimage files are "
+ "created in the same directory as FILE is in by default. \n"
+ "-t [n] [m] set the high and low threshold values. n and m are values between 0 and 1.");
System.exit(1);
}
// interpret options and read arguments
if (args[0].equals("-t")) {
threshLow = Float.parseFloat(args[1]);
threshHigh = Float.parseFloat(args[2]);
filename = args[3];
if (args.length == 5) {
dir = args[4];
}
} else {
filename = args[0];
if (args.length == 2) {
dir = args[1];
}
}
// get path to directory file is in
String name = FilenameUtils.removeExtension(filename);
// get image
BufferedImage image = UtilImageIO.loadImage(new File(filename).getAbsolutePath());
if (image == null) {
System.out.println(
"usage: crop-objects [OPTION]... FILE [DIR]\n"
+ "crops detected objects from image FILE and writes their subimages to files. \n"
+ "Can specify the DIR in which to create the files, otherwise subimage files are "
+ "created in the same directory as FILE is in by default. \n"
+ "-t [n] [m] set the high and low threshold values. n and m are values between 0 and 1.");
System.exit(1);
}
minsize = (int) (0.1 * Math.min(image.getHeight(), image.getWidth()));
// find objects in image
// generate candidate contours
ArrayList<List<PointIndex_I32>> objects = new ArrayList<List<PointIndex_I32>>();
List<BufferedImage> results = new ArrayList<BufferedImage>();
List<List<PointIndex_I32>> candidates = new ArrayList<List<PointIndex_I32>>();
ImageFloat32 input = ConvertBufferedImage.convertFromSingle(image, null, ImageFloat32.class);
BufferedImage bw =
ConvertBufferedImage.convertTo(
input, new BufferedImage(image.getWidth(), image.getHeight(), image.getType()));
File binaryfile = new File(name + "_" + "binary.png");
ImageIO.write(bw, "png", binaryfile);
ImageUInt8 binary = new ImageUInt8(input.width, input.height);
// Finds edges inside the image
CannyEdge<ImageFloat32, ImageFloat32> canny =
FactoryEdgeDetectors.canny(
blurRadius, true, dynamicThreshold, ImageFloat32.class, ImageFloat32.class);
canny.process(input, threshLow, threshHigh, binary);
List<Contour> contours = BinaryImageOps.contour(binary, rule, null);
BufferedImage visualBinary = VisualizeBinaryData.renderBinary(binary, null);
File cannyfile = new File(name + "_" + "canny.png");
ImageIO.write(visualBinary, "png", cannyfile);
BufferedImage cannyContour =
VisualizeBinaryData.renderExternal(contours, null, binary.width, binary.height, null);
File cannyContourfile = new File(name + "_" + "contour.png");
ImageIO.write(cannyContour, "png", cannyContourfile);
for (Contour c : contours) {
// Only the external contours are relevant.
List<PointIndex_I32> vertices =
ShapeFittingOps.fitPolygon(c.external, true, toleranceDist, toleranceAngle, 100);
candidates.add(vertices);
}
for (List<PointIndex_I32> vertices : candidates) {
try {
Candidate c = new Candidate(vertices, image);
if (c.size(minsize)) {
c.rotate();
results.add(c.getImage());
objects.add(vertices);
}
} catch (Exception e) {
System.out.println("Error creating candidate from contour " + e.getMessage());
}
}
// write subimages of objects to files
int i = 0;
for (BufferedImage obj : results) {
// print images to file
try {
File outputfile = new File(name + "_" + i + ".png");
i++;
ImageIO.write(obj, "png", outputfile);
} catch (IOException e) {
System.out.println("Error writing subimages" + e.getMessage());
}
}
// draw objects onto original image and save
Draw.drawPolygons(objects, image);
File outputfile = new File(name + "_" + "annotated.png");
ImageIO.write(image, "png", outputfile);
}
public <II extends ImageSingleBand> double[][] harder(BufferedImage image) {
MultiSpectral<ImageFloat32> colorImage =
ConvertBufferedImage.convertFromMulti(image, null, true, ImageFloat32.class);
// convert the color image to greyscale
ImageFloat32 greyscaleImage =
ConvertImage.average((MultiSpectral<ImageFloat32>) colorImage, null);
// SURF works off of integral images
Class<II> integralType = GIntegralImageOps.getIntegralType(ImageFloat32.class);
// define the feature detection algorithm
NonMaxSuppression extractor =
FactoryFeatureExtractor.nonmax(new ConfigExtract(2, detectThreshold, 5, true));
FastHessianFeatureDetector<II> detector =
new FastHessianFeatureDetector<II>(extractor, maxFeaturesPerScale, 2, 9, 4, 4);
// estimate orientation
OrientationIntegral<II> orientation = FactoryOrientationAlgs.sliding_ii(null, integralType);
DescribePointSurf<II> descriptor =
FactoryDescribePointAlgs.<II>surfStability(null, integralType);
// compute the integral image of the greyscale 'image'
II integralgrey =
GeneralizedImageOps.createSingleBand(
integralType, greyscaleImage.width, greyscaleImage.height);
GIntegralImageOps.transform(greyscaleImage, integralgrey);
// detect fast hessian features
detector.detect(integralgrey);
// === This is the point were the code starts deviating from the standard SURF! ===
// tell algorithms which image to process
orientation.setImage(integralgrey);
List<ScalePoint> points = detector.getFoundPoints();
double[][] descriptions = new double[points.size()][3 * descriptor.getDescriptionLength()];
double[] angles = new double[points.size()];
int l = 0;
for (ScalePoint p : points) {
orientation.setScale(p.scale);
angles[l] = orientation.compute(p.x, p.y);
l++;
}
for (int i = 0; i < 3; i++) {
// check if it is actually a greyscale image, take always the 1st band!
ImageFloat32 colorImageBand = null;
if (colorImage.getNumBands() == 1) {
colorImageBand = colorImage.getBand(0);
} else {
colorImageBand = colorImage.getBand(i);
}
// compute the integral image of the i-th band of the color 'image'
II integralband =
GeneralizedImageOps.createSingleBand(
integralType, colorImageBand.width, colorImageBand.height);
GIntegralImageOps.transform(colorImageBand, integralband);
// tell algorithms which image to process
// orientation.setImage(integralband);
descriptor.setImage(integralband);
int j = 0;
for (ScalePoint p : points) {
// estimate orientation
// orientation.setScale(p.scale);
// double angle = orientation.compute(p.x, p.y);
// extract the SURF description for this region
SurfFeature desc = descriptor.createDescription();
descriptor.describe(p.x, p.y, angles[j], p.scale, (TupleDesc_F64) desc);
double[] banddesc = desc.getValue();
if (perBandNormalization) {
banddesc = Normalization.normalizeL2(banddesc);
}
for (int k = 0; k < SURFLength; k++) {
descriptions[j][i * SURFLength + k] = banddesc[k];
}
j++;
}
}
return descriptions;
}