Research On The Technique Of 3D Human Face Reconstruction Based On Photos

Currently, synthetical human face technique, which is in the area of computerimitating, is now receiving more and more attention than before because of its extensivelyapplied foreground. It can be used for the research of the human body language feeling'smodel, virtual environment, the correspondence technique, extended education, medicaltreatment research, movie making and the game amusement etc. Generating realistic 3D human face models has been a persistent challenger incomputer graphics. As we know, human face shows three-dimension way. But in therealistic life, the image for human face identifying is based on dual-dimension spatialinformation. This exits some limitation. It shows three-dimension object through adual-dimension way, so it is inevitable that a lot of spatial information can not be used. Itmakes human identifying to face some difficulty. According to this kind of viewpoint, inthe thesis, I put forward to identifying the three-dimension human face that has beenrebuilt. The advantage of this kind of method is that it still can be identified after the imageis revolved. And the accurate rate is high. Moreover, after the 3D human face is rebuilt, theimage is the vector graph. So, it can be arbitrarily constringed and it can not take place theform. I am mainly engaged in the research of the rebuilding of the 3D human face in mythesis. It is introduced which is the applied foreground of the rebuilding of 3D human face,the research meaning and the present conditions of domestic and international research.And I carried on the analysis to the existing techniques of this area and the difficulty toface. After studying the knowledge of the stereoscopic sense of vision, the rebuilding of 3Dgraph, and the technique of the human face identifying and putting the value with thespline, I carried on the research of the rebuilding of the 3D human face. It is mainly analyzed about the rebuilding of the 3D human face. The main research iswithdrawing the data of the facial characteristic from the human face photograph. It ismainly making use of the principle of the stereoscopic sense of vision. It adopts thefunction of spline to turn the image to the vector graph. And the Laplace equation is usedto divide the boundary of the face area which is the monotonous district. Then we canmatch the data of the face characteristic in different district. After that, we can calculate thedepth information of the 3D human face. And then, the function of spline is adopted to putthe value into the model. Finally, It is loaded in the OpenGL to display the realistic 3Dhuman face. In the thesis, I provided parts of program code also. This thesis also analyzes the current three-dimension visualization software, OpenGL,and the feasibility of using OpenGL as three-dimension human face visualization software.Then the basic principles and methods of OpenGL are described and the implementationmethod of OpenGL in three-dimension human face visualization is presented. Moreover, it is still pointed out that the JPEG image has the periodic characteristic. Ianalyze the reason from JPEG's compression process, the influence and put forward someimproving suggestion. It is pointed out that when we identify some object with the JPEGimage, the periodic image will bring some trouble. Because of the exploitation of theperiodic characteristic to identifying the non-periodic characteristic, the result is not idealor nonsense. It combines to the experiment result to prove this phenomenon.