A quite simple method to determine clouds shadows is proposed in the paper: '''A Simple Method of Determining Cloud-Masks and Cloud-Shadow-Masks From Satellite Imagery' by Cucu-Dumitrescu, Geoscience and Remote Sensing Letters, IEEE, 2013 (DOI: 10.1109/LGRS.2013.2244842). For shadows caused by elements of urban landscape, I expect that it is a little bit more complicated. Anyway, if you cannot control the satellite overpass time, at least (provided it is possible) you can try to use images acquired from near-nadir orbits.
Thank you Ruxandra for your answer. But what I want is to retrieve the information lost by shadows due to the urban features such as building or trees. I believe that the second part of your answer is more relevant to the question. In addition, I think there should be other techniques where I can retrieve information from other images to replace shadows do you agree?
Mr. Awad, I agree that it is possible to generally retrieve hidden information (e.g., by shadows) processing image time series. Nonetheless, given the fact that satellite overpass hours are always the same, the shadows caused by urban features are spatially correlated. In this situation, you should try to select images acquired while the solar elevation is as different as possible. As regard Mr. Aftab's reply, I must acknowledge that I have no experience in the tool he is recommending. However, I expect that hidden information retrieval depends heavily on the input images selection. Complementary to near-nadir viewing, I would select them taking into account the geometry of acquisition (relative position between Sun and the satellite).
I presented a post-prosseing step for removing shadows on rooftops. To gain more information, you can refer to the following publication:
" Automatic building extraction in dense urban areas through GeoEye multispectral imagery "
- Badges
- Science topic
- Similar topics
- Engineering
- Electrical Engineering
- Signal Processing
- Image Processing