Evolutionary Active Contours for Muscle Recognition

Active Contours constitute a widely used Pattern Recognition technique, consisting in curves to be moved by minimizing an energy function. Classical active contours are based on different methodologies. This paper proposes a new procedure using Genetic Algorithm to minimize the energy function of the Active Contour. Both techniques are combined to produce a new methodology, called Evolutionary Active Contours. An application has been developed in order to verify the validation of the proposed method. The valid performance of this approach has been demonstrated with both synthetic and real Magnetic Resonance images. Moreover, the algorithm has been used to recognize muscles from Magnetic Resonance images of Iberian ham at different maturation stages in order to calculate their volume change. Our main findings can be summarized as two: The feasible combination of genetic algorithms and active contours, and the automation of optimal ripening points for Iberian ham samples.