This is an example project, showcasing how you can use the ‘pixel-based object resizing’ algorithm to generalize your image objects. For this algorithm you can choose between 3 different resizing modes: Growing, shrinking and coating.

• Growing means one row of pixel is added an merged with the original object per execution.
• Shrinking means one row of pixels is taken away from the original object and classified in a separate class.
• Coating means one row of pixels is cut around the original object and classified to a separate class.

With the algorithm ‘pixel-based resizing’ conditions can be set for pixels to be added or cut.

Pixel constraints per layer:
You can set pixel layer constraints, e.g. the pixel to be grown into must have low values for panchromatic layer.
Surface tension:
You can define surface tension per object or per class. A defined box is then drawn for every pixel and the entered operation and value represent to coverage of the object/class. If the surface tension condition is fulfilled the object will then grow or shrink, if not, nothing is executed.

Applications

• Generalize object outlines to get a smoother vector layer output.

• Create a buffer around objects, i.e. because you want to cut a so called region from it.

• Create square objects and grow them exactly pixel wise (part of customized algorithm downloadable from the knowledge base).

Please find here the example project for object generalization of complex Landcover. In the example a surface tension of <0.5 and >0.5 is set to grow and shrink an object.

Download 'Example Project - Pixel-Based Object Resizing Object Generalization Complex Landcover'

Interesting related content:

• eCognition Deconstructed video: Pixel-based Object Resizing I
• eCognition Deconstructed video: Pixel-based Object Resizing II
• Project & Rule Set examples: Smooth Image Objects
• Project & Rule Set examples: Example Project - Pixel-Based Object Resizing Surface Tension Artificial
• Project & Rule Set examples: Building Generalization