Showing posts with label Sobel Edge operator Computer vision. Show all posts
Showing posts with label Sobel Edge operator Computer vision. Show all posts

Sunday 26 December 2010

Image algorithms demo

At last an update to my blog! This has been a busy year for me with my work, I am now a developer in Helse Midt-Norge IT since May, earlier this year I worked as an independent consultant for Artsdatabanken (NTNU). I work with WPF and C# mainly presently.

I will now present a new hobby project of mine.

This WPF application demo of image algorithms demonstrates different techniques towards
blur, sharpen, edge detection and other filtering techniques, all relying on the general concept of Convolution in image processing. 8-bits level grayscale images (256 levels, 1 bpp) are supported.




I have worked with an image algorithms demo the last days. The
demo is a small WPF application that demonstrates image algorithms used in
edge detection. See the link below to download the zip file that contains
the Microsoft Visual Studio 2010 project with source code.

Examples of edge operators are sobel edge operator and Prewitt, Laplacian, Roberts and
Horizontal and Vertical Sobel, blur difference operator, shown above. There are additional
filters included.

The resulting images can be inverted. The application works with 8-bit grayscale images (256 levels). The BitmapSource class must be able to import the image (BMP or GIF should work).

Right now the demo is in v1.1. I will add more features soon. There are a few issues with double-edges, noise and normalization in the demo, but I hope I will complete the demo in a more stable version later. Feel free to unzip the zip file linked below and provide comments of the code. Perhaps you can add new features to my code and discuss optimizations, corrections?

I will state that already there is very little code to make a new filter, so this source code could be used as a basis for developing filters for applications needing to provide basic blur, sharpen and edge detection capabilities.



Sobel operator, horisontal komponent (3x3 convolution kernel).

Comment:

Download newest version of ImageAlgorithmsDemo WPF Application here [v1.1.]

Image algorithms v1.1 (Zip-fil, ca. 2,4 MB compressed)

Last update 28th of December 2010.

Future stuff:
  • improve the speed of the filtering
  • support greater than 8 bpp (e.g. RGB)
  • support more operators. Canny edge and more advanced filters are desired
  • improve code, add tests to the test project
  • even friendlier GUI, more feedback and information
  • document code
  • share code with the community when v2.0 is done.
The goal is that this demo should be a good starter demo of edge analysis in digital image algorithms, that can be used in courses at universities and the likes. The demo demonstrates pixel manipulation in WPF-application and filtering also with up to date .NET 4.0 code. Often one only finds C or C++ code when looking for these kinds of algorithms..

Comments?