Algorithm – Sugarcube

September 5, 2022

ImaproFX – Image Processing Framework

ImaproFX – JavaFX Image Processing Framework

Grasping the essence of image processing and analysis algorithms may prove to be quite difficult. To simplify the comprehension of such algorithms, sugarcubeIT proposes ImaproFX, a Java FX user interface providing live visual feedback thanks to interactive controllers.

The real power of ImaproFX comes from it’s Java coding framework… Coders can create their own image analysis ribbons simply by extending and implementing an abstract “ImaproFilterFx” class :

public class ContrastFilterFx extends ImaproFilterFx
{

    public @FXML Slider contrastSlider, luminositySlider;
    public @FXML CheckBox invertCheckbox;

    public ContrastFilterFx(FxEnvironment env)
    {
        super(env, "Contrast & Luminosity");
    }

    @Override
    public void reset()
    {
        contrastSlider.setValue(0);
        luminositySlider.setValue(0);
        invertCheckbox.setSelected(false);
    }

    @Override
    public void process(FilterImage image)
    {
        double contrast = Math.tan((0.5 + contrastSlider.getValue() / 200) * Math.PI / 2.0);
        double luminosity = luminositySlider.getValue() / 100;

        if (invertCheckbox.isSelected())
            contrast = -contrast;

        for (int y = 0; y < image.getHeight(); y++)
            for (int x = 0; x < image.getWidth(); x++)
            {
                float value = (float) ((image.getValue(x, y) - 0.5f) * contrast + 0.5f);

                if (value > 1)
                    value = 1;
                else if (value < 0)
                    value = 0;

                image.setValue(x, y, value + luminosity);
            }
    }
}

public class ContrastFilterFx extends ImaproFilterFx

{

public @FXML Slider contrastSlider, luminositySlider;

public @FXML CheckBox invertCheckbox;

public ContrastFilterFx(FxEnvironment env)

{

super(env, "Contrast & Luminosity");

}

@Override

public void reset()

{

contrastSlider.setValue(0);

luminositySlider.setValue(0);

invertCheckbox.setSelected(false);

}

@Override

public void process(FilterImage image)

{

double contrast = Math.tan((0.5 + contrastSlider.getValue() / 200) * Math.PI / 2.0);

double luminosity = luminositySlider.getValue() / 100;

if (invertCheckbox.isSelected())

contrast = -contrast;

for (int y = 0; y < image.getHeight(); y++)

for (int x = 0; x < image.getWidth(); x++)

{

float value = (float) ((image.getValue(x, y) - 0.5f) * contrast + 0.5f);

if (value > 1)

value = 1;

else if (value < 0)

value = 0;

image.setValue(x, y, value + luminosity);

}

Post written by Jean-Luc Bloechle

June 29, 2022June 29, 2022

YinYang – Document Image Binarization

The binarization algorithm proceeds in 5 main steps

Have a try at SugarCube’s YinYang online demo !

Here are some YinYang binarization samples applied to difficult document images.

April 19, 2021May 14, 2021

Deep Neural Network – Learning to draw

An interesting feature of Deep Neural Networks is their ability to tend to a simplified image using few neurons and many layers… Watch this CNN training evolution giving birth to AI artistic painting !

Using sugarcube’s Deep Learning API :

CNN cnn = new CNN(1, 1, xy.length);
for (int i = 0; i< 11; i++)
  cnn.addFull(16, CNN.RELU);
cnn.addRegression(3);

CNNTrainer trainer = cnn.trainer();
trainer.method = CNNTrainer.SGD;
trainer.learningRate = 0.01;
trainer.momentum = 0.3;
trainer.batchSize = 1;
trainer.l2Decay = 0;
trainer.reset();

CNN cnn = new CNN(1, 1, xy.length);

for (int i = 0; i< 11; i++)

cnn.addFull(16, CNN.RELU);

cnn.addRegression(3);

CNNTrainer trainer = cnn.trainer();

trainer.method = CNNTrainer.SGD;

trainer.learningRate = 0.01;

trainer.momentum = 0.3;

trainer.batchSize = 1;

trainer.l2Decay = 0;

trainer.reset();

Post written by Jean-Luc Bloechle

March 6, 2021May 14, 2021

Algolive – Sorting algorithms visualisation

Understanding sorting algorithms may be a bit tricky : InsertionSort, SelectionSort, BubbleSort, ShellSort, QuickSort, MergeSort.

In order to simplify the comprehension of such algorithms, sugarcubeIT propose Algolive, a simple animated user interface implemented in HTML5 (CSS3 & JavaScript) able to visually decipher the key steps of several major sorting algorithms.

AlgoLive is freely accessible for any personnal and teaching use : www.sugarcube.ch/algolive

Post written by Jean-Luc Bloechle

September 16, 2016March 14, 2021

Scan Background Removal

Removing the paper background from scanned document is not a trivial task. Here is how sugarcube addresses the problem.

Objectives

Remove paper background from scanned documents in order to create searchable PDF files composed of binary images together with transparent text layers
Binary images assume black text over white background in order to get :
- a good reading experience
- a compact file size
Develop an automatic process to apply on a batch of tif files

Facts

420’000 TIF images, scanned pages from “Recueil des lois fédérales” from 1947 to 1998 (german, french and italian versions)
a total amound of 6,8 TB (TeraBytes)
image resolution : 300 DPI, 24-bits RGB

Results

Here above are some representative input samples of the corpus with their respective output counterparts :

How-to

First, we get a TIFF image from the scanned Swiss “Bundesarchiv”.
Our algorithm then computes the mean background colors for small image tiles.
The resulting blocky effect is filtered using a bilinear interpolation.
The algorithm subtracts the background image from the original one, resulting in a non homogeneous light background.
A final dynamic gamma correction is applied to get rid of remaining artefacts.

Conclusion

Getting rid of scanned document paper background is not a straightforward process. Our experience shows that scanned documents variability forces the implementation of an adaptive algorithm.

For instance, tuning a binary threshold from grey level images is clearly not a viable solution with such a heterogenous corpus containing images subject to luminosity, contrast and hue defects.