ARDECO: Automatic Region DEtection and COnversion

Vector Image Galleries

 

Perceptually-based rendering

We show here the versatility of Ardeco, that can take into account perceptual data. Here, importance is computed from eye-tracker data (available on the web in De Carlo et. al's work). As in De Carlo et. al's paper, details in the background of the image are simplified (see for instance the shirt of the sitting man). Note how gradients and SVG Gaussian blur further improve the sensation of depth and reproduces the shading of the original image.
If vector images are not displayed, please use this link instead or install the SVG plugin.
 

Original bitmap image
Importance computed from eye-tracker data.

Saliency map.
Trixels adapted to saliency and importance.

PDF
Ardeco approximation with constant colors.
Blurring unsupported in PDF
Approximation with constant colors and background blur.

PDF
Approximation with gradients.
Blurring unsupported in PDF
Approximation with gradients and background blur.

Original bitmap image
Importance computed from eye-tracker data.

Trixels
Blurring unsupported in PDF
Approximation with gradients and background blur.