Reputation: 325
The image shows 4 images.
Image2 has been co-registered as per Image1, so that change detection can be performed. This results in a black portion in the right side and bottom side of the aligned image(3rd one), because that portion is not available in Image2. Or we can say,there is a loss of information in aligned image,when compared to Image2(due to registration).
4th image shows the detected changes in white on a black background. As we can see,in the right and bottom portions of 4th image,some change has been detected(white regions).
As per my requirement,these changes need not be reported. The reason is Image1 contains more information than Image2,which is being shown by black portions in aligned image(3rd). Therefore that extra information need not be qualified as change.
The area blackened by co-registration should be neglected by my change detection algorithm.
How can I achieve that? I am looking for ideas to detect such blackened areas.
Upvotes: 0
Views: 418
Reputation: 15867
If you can transform a uniform white image using the same parameters, you'd have a b/w mask of "reachable" pixels, that you can e.g. AND with the change detection result
Upvotes: 2