Edges Extraction for Robotic Application


This page illustrates the application of Becheur paradigm to image segmentation in order to extract the edges of an image of a solid (our goal).

The different from the previous solution is related to the detetion of the markers. In this approach a more automatic method is used, based on regional minima and dynamics.

The input image is shown bellow.

fig. 1 - Original Image

The image is first filtered by a 3x3 median filtering:

_fig. 2 - Median filtering

In order to extract the edges it is necessary first to take the Morphological Gradient of the image. That is shown in the image bellow (the image is inverted to better show all details).

fig. 2 - Gradient

The markers are found using regional minima of the gradient which have a minimal dynamic of 10 and a minimum area of 500 pixels.

fig. 3- Markers (labeled for better visualization)

The next step in the paradigm is to apply another operator known as minimum imposing transformation and this result is segmented by the watershed operator. The result is shown below (this image is also inverted).

fig. - Gradient with minimum imposing on the markers

Now we can apply the Watershed. The result is shown below.

fig. 6 - Watershed operator result


workspace: segblock-min.wk

Back to the initial page