Research And Implementation Of Eye-tracking Autostereoscopic Display Technology Based On Rearranging RGB Sub-pixels

Autostereoscopic display technology makes the traditional stereoscopic display out of the auxiliary equipment of the shackles.Which ridge-type's development is the most rapid,but its shortcomings is the best point of view fixed and discontinuous,the viewer after the head moves from the best point of view will appear crosstalk or even left and right eye image reversal.To solve the problems above,this paper proposes an eye-tracking autostereo-scopic display method based on rearranging RGB sub-pixels.It focuses on the research and implementation of human eye-tracking module,and implements the stereo video playback system in Windows and Android system.The main contents are:1.The binocular stereoscopic vision model is studied by human eye stereoscopic vision principle.The principle of autostereoscopic display of grating is introduced.The causes of crosstalk are analyzed emphatically.And then put forward the image compensation technology of the tracking autostereoscopic display under the condition of tilting placed grating.The core of the technology is to track the viewer position in real time,change the sub-pixel arrangement of the stereo image,and then realize the change of the viewpoint in the spatial position,which can reduce the maximum crosstalk to 16.7%.2.Depth study and implementation of human eye-tracking module.Firstly,face detection based on the Haar classifier,including the Haar-like feature,the fast calculation principle of the integral graph,the related theory of the AdaBoost algorithm and the face detection process based on the AdaBoost concatenator.Secondly,eye detection based on the ASM algorithm,including the shape statistical model,the establishment of the local texture model and the search matching of the feature points.Finally,algorithm improvement,the introduction of skin color detection and template matching,based on OpenCV to implement eye-tracking module,in improving the detection accuracy while also accelerating the detection speed.3.Integrated tracking autostereoscopic display system on Windows,combined with the human eye-tracking module and video playback module to achieve three-dimensional video playback system.The correspondence between human eye position's coordinates and subpixel arrangement is studied emphatically,and the measurement method of correspondence table in engineering practice is introduced.Produced a 32-inch tracking autostereoscopic display and analyzes the overall performance.Verify the validity of the image compensation method proposed in this paper.4.The stereo video playback system is transplanted to the Android system,through the JNI mechanism to transplant FFmpeg multimedia framework to solve the real-time video frame fusion,focusing on the FFmpeg decoding process.A 10-inch tracked free stereoscopic display is produced and the overall effect of the system is demonstrated.