THREE DIMENSIONAL FACE SHAPE RECONFIGURATION METHOD CAPABLE OF AUTOMATICALLY PRODUCING A FACE OF A SUSPECT WHICH IS RECORDED IN A CCTV CAMERA TO A THREE DIMENSIONAL SHAPE

摘要

PURPOSE: A three dimensional face shape reconfiguration method is provided to restore a three dimensional shape of a face of a person with high definition from a moving picture which includes a front face image of the person.;CONSTITUTION: A target frame among input frames is designated to an apparatus and the apparatus recognizes a face area of each frame and one or more characteristics partial areas within the face area (S102,S103). The apparatus warps the rest frames and magnifies the target frame and face areas of the rest frames to the size corresponding to high definition (S104,S105). The apparatus divides the magnified target frame and the face areas of the rest frames into multiple unit patches and changes all pixels forming the face area of the target frame based on the similarity between the unit patches and obtains a high definition face image (S106-S108).;COPYRIGHT KIPO 2013;[Reference numerals] (AA) Start; (BB) End; (S101) Press successive multiple frames; (S102) Designate a target frame among input frames; (S103) Recognize a face area of each frame and one or more characteristics partial areas within the face area; (S104) Warp the rest frames corresponding to the rest frames; (S105) Magnify the target frame and face areas of the rest frames to the size corresponding to high definition; (S106) Divide the magnified target frame and the face areas of the rest frames into multiple unit patches; (S107) Calculate the similarity between the unit patch frames and the other frames; (S108) Change all pixels forming the face area of the target frame based on the similarity between the unit patches and obtain a high definition face image; (S109) Reconstruct a three-dimensional face shape based on the obtained high resolution face image