There is plenty for you to tackle in this field. This at least is one conclusion from the papers in this handbook. They describe the active areas of investigation. Despite the massive increase in computational power available to reseachers since the 80s, and better algorithms, pattern recognition still faces unknowns.

Closely tied to this is the entire issue of what constitutes robust computer vision. The book gives examples of applications where better methods would have tangible uses. Like being able to identify objects in a video stream. Or recognising human faces. The latter problem has achieved some prominence due to a heightened anti-terrorist mood since 2001. So at least some problems can usually get the researcher adequate funding.