Generalized Hough Transformation


This applet demonstrates the Hough transformation for general shapes. It detects any shape in the image space by finding clusters in the Hough space.

How to use the applet

Below you can see three images, two big ones and a small one. The first big image on the left is the input image. This is the image where all shapes are searched in. The image on the right side shows the Hough space. You can detect probable shapes in the input image by finding clusters in the Hough space. The small image shows the active template. The shape that can be seen in this image will be searched. The red point on the upper left of the template image is the reference point of the Hough transformation. To change the template image use the -button.When the applet starts just a black picture is loaded. You can draw your own shapes in this image by dragging the mouse pointer over it. If you do so observe the Hough space and see the growing maxima. If you want to delete your work press the -button. When you just move the mouse over the input image a blue rectangle with the template shape in it appears on the right picture. This is the Hough transformation of the point that would be add in the Hough space if you draw a dot by pressing the mouse button now.

To choose other input images with alternative shapes press the -button. The mouse pointer over the Hough space produces a red rectangle on the left image. This is the assumed position of the template shape. You can instruct the applet to find some clusters automatically by pressing the -button. Then you are requested to choose the number of maxima that should be found in the input image. The positions of the clusters in the Hough space are marked by green rectangles whereas the assumed positions of the shapes in the input image are marked by red rectangles. To let them disappear just press the left mouse button.





Authors: René Iser, Behrang Karimibabak and Simon Winkelbach

Other Applets