Could you please tell me how to solve my problem with segmentation in ERDAS ?

I am using ERDAS 2013 and the tool Image Segmentation to segment my multiband picture (with high resolution).

Unfortunately, the results show the mosaic tiles, that come as a result of incorrect color matching in the mosiac dataset.

I do not have the original raster catalog to redo the color matching. How should I perform my segmentation to get rid of the visible tiles on the segmentation reult?

See the image: enter image description here

  • If that's a lake in the lower-right, it might be that the segmentation is in fact working correctly, and the input data has dummy values there. The red blocks look a lot like the whole tiles that don't touch any shore. – Paulo Raposo Nov 3 '15 at 5:40

Image segmentation works by applying any of a number of algorithms to group similar pixels into objects. Most image analysis does not end with segmentation. Rather, you need to add another step in you workflow to classify these objects into usable information classes. One method for reclassifying these large tiles in the image would be to set up a size filter, so that any objects > x are reassigned a value of 0, for example. There is a good tutorial video that can walk you through the process of image segmentation and classification in Erdas Imagine Objective.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.