Cartoon face synthesis based on Markov Network

In this paper, an effective method for face photo-cartoon synthesis is used based on Markov Network model. The basic idea of this method is to synthesize a cartoon drawing from the database based on a cartoon drawn by an artist. This method consists of three steps. We first manually labeled eyes at fixed position to ensure the structure of the face and the cartoon face alignment roughly. Second, those faces and cartoons are divided into small overlapped patches, which can generate photo-cartoon pairs in the sample set. We build the joint photo-cartoon model based on the Markov Network model finally. Compared with the existing approaches, our method is simple to implement and more similar. We evaluate our approach on CUHK student database and demonstrate good performance.