DO-IT-YOURSELF PHOTO REALISTIC TALKING HEAD CREATION

摘要

A method and an apparatus for constructing a photo realistic human talking head are disclosed. The apparatus comprises a handheld device; a network and a server. The server comprises the library of alignment templates, a controller, a mixer and a storage means and the handheld device comprises a video camera and a display. The network and the handheld device communicate one with the other and the network and the server communicate one with the other. Each alignment template is different one from the other and representational of an alignment template frame of a photo realistic human talking head having an alignment template head angular position comprising a template head tilt, a template head nod, and a template head swivel component. Each alignment template frame is different one from the other and each alignment template head angular position is different one from the other. The controller selects an alignment template from the library of alignment templates corresponding to a specified alignment template head angular position and having an image output signal representational of the alignment template. The video camera collects an image of a human subject having a head having a human subject head angular position comprising a human subject head tilt, a human subject head nod, and a human subject head swivel component. The video camera has an output signal representational of the collected image of the human subject. The handheld device communicates the output signal of the video camera representational of the collected image of the human subject to the server via the network. The server has an input receiving the output signal of the video camera representational of the collected image of the human subject. The server receives the selected alignment template image output signal from the controller, and communicates the selected alignment template image output signal and the received collected image signal of the human subject to the mixer. The mixer receives the selected alignment template image output signal and the communicated collected image signal of the human subject, and mixes one with the other into an output signal representational of a composite image of the collected image of the human subject and the selected alignment template, and communicates the composite image signal of the collected image of the human subject and the selected alignment template to the server. The server has an output signal representational of the composite image signal of the collected image of the human subject and the selected alignment template received from the mixer. The server communicates the output signal representational of the composite image signal of the collected image of the human subject and the selected alignment template to the handheld device via the network. The display has an input receiving the output signal representational of the composite image signal of the collected image of the human subject and the selected alignment template. The display and the video camera are adapted to allow the video camera to collect the image of the human subject and the human subject to view the composite image and the human subject to align the image of the head of the human subject with the alignment template head at substantially the same angular position as the specified alignment template head angular position. The storage means stores a library of collected images of the human subject with the head of the subject at different human subject head angular positions. The server communicates the received collected image signal of the human subject to the storage means. The storage means receives and stores the received collected image signal of the human subject as a stored image of the human subject when the human subject has the head of the human subject substantially aligned with the alignment template head. The stored image of the human subject has the human subject head angular position substantially the same as the specified alignment template head angular position. Each stored image in the library of collected images is different one from the other and each stored image human subject head angular position is different one from the other. Each stored image human subject head angular position of the library of collected images corresponds to and substantially the same as and aligned with a selected the alignment template head angular position of the library of alignment templates and each stored image represents a different frame of a photo realistic human talking head.