Improving Interactive Differential Evolution for Cartoon Face Image Combination

In the general cartoon face methods, these methods can generate realistic cartoon face images, but rarely consider the subjectivity of users. In this condition, algorithms lack direct user intervention or the user operation method is not simple enough, so they cannot meet the needs of different users. In this paper, a method of cartoon face generation based on interactive differential evolution is introduced to overcome the deficiencies of general cartoon face methods. In the new proposed oppositional-mutual learning interactive differential evolution algorithm, named OMIDE, the two strategies of dynamic oppositional learning and mutual learning are combined to reduce the number of generations of algorithm. In addition, the user's dynamic selection is introduced to solve the problem that users are not satisfied with some parts of the cartoon face image while others are satisfied in the later stages of the algorithm. The generation of the proposed method is extensively evaluated by comparisons between it and other methods. In addition, the method of cartoon face generation based on interactive differential evolution is confirmed by a set of experiments to be able to satisfy the needs of different users.