Research On The Technology Of Cartoon Face Stylization Based On Augmented Reality

With the rapid development and progress of information science technology，digital entertainment has been gradually integrated into people’s lives. Technologiesfor Non-photorealistic face and cartoon-like face generation have become researchtopics in the field of computer graphics and computer animation. And they areapplied in online games, mobile digital entertainment, and so on. Therefore, the studyof cartoon face stylization is of great significance and practical application value.This paper presents a new approach to realize cartoon face stylization.According to the fact that there are similar faces if the number of faces increases,and similar faces can share the same cartoon face, this paper establishes a databasewhich contains typical faces ande their cartoon figures. We get the cartoon figure ofthe input face by similarity matching, then use the cartoon figure replace the headregion of the input face by combining the matting and image fusion. With the imagesynthesis, we get painted cartoon face stylization based on augmented reality.Compared with image filtering or characteristics deformation cartoonish method, ourmethod not only produces more vivid cartoon without losing image information, butalso has the effect of cartoon comedy.According to the established database of human face, we draw the correspondingcartoons to create the cartoon gallery by vector drawing tools. On facial similarmatching, this experiment shows that the recognition rate of the PCA algorithm basedon image Euclidean distance is higher than the PCA algorithm based on Eulideandistance, because image Euclidean distance takes full account of the spatialrelationship of pixels in order to avoid slight deformation effect of the image onrecognition results. We use closed-form solution to natural image matting whichneeds user places scribbles to mark foreground and background. With this method,we can get a new background image without person’s head region for integration withthe cartoon figure. And this approach satisfies our system’s requirements better. Toachieve cartoon figure with a new background image seamless integration, we use Poisson Image Editing to realize image synthesis. With this, we finally achievedcartoon face stylization based on augmented reality. We could see that the finalgenerated image has more effects of comedy.