Summary: | 碩士 === 國立臺北科技大學 === 資訊工程系研究所 === 104 === This paper provides an automated image synthesis system, allowing users to combine Chinese paintings with their photo, resulting in a personal style like portrait correspondence. Users only need to provide their own photographs, the system automatically analyzes the facial features, and incorporated by reference portrait fusion rendering. Our system ensures that end users do not need any manual operation. At the initialize step, we only need to input some parameters such as coordinates, the mask map, and orientation of painting&;#39;s face. System will reference input information to extract the face feature and transform the image, referring the Chinese painting texture to make the blending target. Then we will make seamless blending with face features, so that color differences can be minimized, the final will match the original input image color information, the results of the completion of a further modified so that the results can be more natural and closer to the original. Using our system can make people easily make China portraits without any drawing ability.
|