Our feature extraction techniques have concentrated on static and on moving shapes, both on arbitrary and on conic sections, in two and in three dimensions. We were the first to develop a new arbitrary feature extraction technique wherein the arbitrary feature was described analytically. We were also the first to develop evidence-gathering approaches for moving shapes (as expected, we moved to moving arbitrary shapes after that!). Recently we have shown the relationship between evidence gathering (the Hough transform used in Computer Vision) and the Principle of Duality (as used in Algebraic Geometry). We were also the first to solve the initialisation problems in snakes with our dual active contour. As such, many of our students have developed new and novel techniques for feature extraction and description. Every one of these techniques has focussed on a real application, and these provide and excellent vehicle to describe why and how we do this work. In biometric applications we identify people by behavioural or physiological characteristic. We were early workers in automatic face recognition where we have developed some of the most sophisticated techniques for finding eyes.