i want to perform a comparison between every two segments in the same location in different date therefore i want the same segmentation on two different date images .
thanks.
Hi Nasser,
You can use the same parameters for both of the segmentations, but as segmentation is combining a group of similar pixels into one object, depending on the difference in the spectral characteristics, you might not get to same segmentation.
It is normal to have a difference, the times and times are different
Hi Nasser,
you can load both images in the same eCognition project. Then, you perform the segmentation taking into account the band weights that you want to use. For example, if you want to use both images for the object creation, you give all the bands a weight of 1. If you only want to adjust the segments to the second image, give a value of 0 to the band weights of the first image. After this you will have the same object limits superimpose to both images.
Another option is doing the segmentation in one image and then export it as a vector layer.
That will create one segmentation from the two images, i suppose Nasser wants to segment both images separately and then compare them.
By the way, in my opinion when one wants to keep the area of change detection same, then he should consider grid-based analysis after undertaking object-based classification of both images.
1. as suggested above, use segments of one image and then use the same segments to extract feature from other image - and do the change anlaysis after classification of both images separately - the scale of analysis will be same in this case ie an object changed from one class to another or not...
2. load both images earlier (eg 2001) and later(eg 2010 ) - then first segment based on the earlier image's bands and classify , followed by another segmentation using the later image' bands but at a finer scale (means smaller segments) and then classify...Thus, you can find changes at sub-object level or changes in an object...
Note : You may also be supposed to keep the decision rule same, if you use sample based then option 2 will become trickier and you need to take new samples inside the previous sampling objects
it depends your desired analysis scale which option you want to use, then may be other options
- Badges
- Science topic
- Similar topics
- Analytical Chemistry
- Extraction