PDF version of this document
If this research reaches and obtains its goals, more theoretical and practical grounds will be available for future applications of non-rigid registration in active appearance models and vice versa. Such goals are quite important as appearance models have not yet become as wide-spread as one could argue they should be. By making real use of the intensity information that is available in appearance models, an exceptional practical strength is exploited. Full image synthesis is an application where no other model type can yet be seen as a substitute. With more automation in place, higher accuracy, compatibility with other technologies, awareness and ultimately, ubiquitous use of the technique can be shortly anticipated.
It is worth pointing out that availability of active appearance models to individuals who deal with registration of images would take this technology one step ahead. This will introduce more concepts, metrics and studies which increase their functionality and flexibility. Reciprocally, previously ``foreign'' techniques can extend the functionality of active appearance (and maybe shape) models once they are put in the hands of groups with difference background and expertise.
As an instance for the first of the two contributions above, a radiologist could very comfortably view a highlighted model of an organ that is deformed in a natural and sound manner. Results and analysis can be managed rather quickly and neatly as automation and synthesis generation should be made operable and even interactive43.