Construction of human faces from textual descriptions

The FAce SYnthesis (FASY) system described in this paper, presents a novel face construction approach based on textual description with the stored facial components extracted from different face databases. The system has two types of databases: (a) Full Face Database (DB-F) consisting of frontal face images collected from different face databases (b) Facial Component Database (DB-C) consisting of facial components extracted from the faces of DB-F. Both the databases also contain the textual description of the face images/facial component images. If the desired face as per the description of the user is not available in DB-F, then new face is constructed with the help of DB-C. The experiment has been conducted with 200 male and female face images from different face databases. The successful extraction of the facial components from the face images as mentioned above has been found to be 93% on an average and face construction satisfies the user’s textual query in 80% cases on an average. The work has been done using Visual Basic 6.0 and Matlab 6.5. The face construction method is implemented in VHDL, simulated by Modelsim SE/PE, and synthesized with Xilinx Webpack 4.1 followed by loading into the FPGA device.