A System for Generation of Face Images from Textual Descriptions

Face generation plays an important role in designing video games, virtual reality and other entertainment applications. Also face image generation from the description of eye-witness is an important step in criminal identification. As per normal human practice, a face is described using textual description involving adjectives and/or some physical properties of its constituent components (the eyes, nose, lips etc.). The standard practise of face image generation is to employ an artist who creates the face images from the textual description. However, an artist with this peculiar expertise is generally rare and the drawn face images may not satisfy the eye-witness/descriptor. This process is very time consuming too. Again the artist draws according to his/her own realization for the textual description and the descriptor, who does not have any experience with paints and brushes can not interact with drawing process. All these limitations of a human artist call for development of an automatic face generation system and this necessity has inspired and initiated research activities in the area of face generation from human-like description and this is the main objective of this work

Santanu Halder received the MCSE and Ph. D.(Eng.) degrees from Jadavpur University, India, in 2005 and 2012 respectively. He is associated with Goverment College of Engineering and Textile Technology, Berhampore in India. His research interests pertain to the applications of computational intelligence techniques like GA, ANN, Cellular Automata etc.