In this topic you have explored methods to enhance and extract information
contained in the temporal domain. You have identified techniques to process
multitemporal images and equipped yourself with the necessary skills.
You should also be able to explain the following statements:
- Information about change is critical to understanding trajectories
of environmental and urban systems.
- A challenge is to isolate those trajectories from the natural variation
within the system.
- Models of temporal variation are relatively immature and there is
scope to strengthen them by revisiting some of the fundamental principles
of data modelling.
- EO images provide only a snapshot in time.
- Our ability to monitor change accurately depends on the frequency
and rate of change relative to the revisit frequency of and spatial
resolution of data acquired by EO missions.
- When choosing a technique to detect change the analyst must be aware
of the type of data s/he is processing.
- It is only possible to compare like with like data sets.
- Some techniques, such as image subtraction, only identify the presence
of change. Others, such as post-classification comparison and change
vector analysis, yield information on the nature of change. And others,
such a principal components analysis, can yield information on the complex
of temporal variations but components first have to be interpreted.
It may be challenging to interpret eigenvectors of multitemporal data,
even for experienced image analysts.
- Lengthening archives of EO data mean that the future for multitemporal
image processing will be very interesting. New, robust statistical techniques
are likely to be made more widely available but it is, I believe, advances
in intelligent methods that are likely to yield the more significant
long-term benefits for image interpretation. Top of the research agenda
is to combine models of extracting information in the spectral, spatial
and temporal domains.