Last updated on 3 May. 2007

Rambler's Top100

Smart Edges RUS ENG

The commercial project made for presentation of new algorithm of scaling of raster images. All rights on application of the given algorithm belong to the author of the project.

 

Resampling

 

Deblurring

 

Dejpeg

 

Smart Edges

 

Smart Picsel

 

New Mosaic

 

Denoise

 

Deconvolution

 

Dequantization

 

Smart Edges - the new method for creation smart edges directions map.

 

In a method described in patent US 6,928,196, all over again using values of elements of the image, the map of edges is designed. Then to a map of edges apply morphological operations. Then a map of edges sorting on four directions: horizontal (0), diagonal (π/4), vertical (π/2), and parallel to a collateral diagonal (3π/4). The map with four possible directions can appear insufficient in case of application of exact algorithms of processing of images.

  Unlike usual adaptive algorithms where the method of calculation of brightness for each element of the formed image depends only on corresponding value in a map of directions of edges, in an offered method for increase in accuracy it is offered to consider and a degree contrast edges. Unlike usual methods of map development of directions the offered method has an opportunity of optimization of parameters of algorithm from the point of view of accuracy and speeds depending on requirements and for each case of realization.

In order to confirm above, I suggest to get familiar with practical realization of the algorithm on the well known examples of tasks and solutions.

As the test the following representation has been chosen:

rings lhouse

To see the picture click on the fragment

 

Test - rings

 

Test - lhouse

Original image was reduced 0.5X using a box convolution kemel.

 

 

Example interpolation with a new map of edges

 

RS-M-spline2 - New updating of algorithm which differs from the previous version (RS-M-Spline) application of a new smart edges directions

map. As a result of introduction in algorithm of adaptive to local orientation of edge the mistake of interpolation or a filtration has essentially decreased.

 

Test - lhouse

Original image was reduced 0.5X using a box convolution Kemel.

 

Before After


 

RMSE 11.52

 

 

Low resolution image was enlarged 2X by variety of methods.

RS-M-Spline (without smart edges direction map) RS-M-Spline2 (with smart edges direction map)


RMSE 9.23 RMSE 8.86

 

 

DJ-spline2 - New updating of algorithm which differs from the previous version (DJ-M-Spline) application of a new smart edges directions

map. As a result of introduction in algorithm of adaptive to local orientation of edge the mistake of interpolation or a filtration has essentially decreased.

Test - lhouse

Original image was reduced 0.5X using a box convolution kemel and has been compressed.

 

Before Atter


RMSE 12.73

 

The compressed image has been restored by means of algorithm DJ-spline and DJ-spline2

The resolution of image will be enlarged by 2 times

 

DJ-M-Spline (without smart edges direction map) DJ-M-Spline2 (with smart edges direction map)


RMSE 11.21 RMSE 10.92

 

The opportunity of realization in other areas is at the moment studied.

 

 
Your comments and questions:
vitalybn@fpy.ru
resampling@yandex.ru
 
The algorithm is constantly improved, improvements will be, wait.

 

Hosted by uCoz