An Algorithm Research And Implementation Of Eye Detection And Tracking For 3D Naked Eye Display

With the vigorous development of 3D display technology,humans put forward higher requirements for convenient and comfortable 3D display.Since the naked-eye 3D display did not need to wear glasses when watching,it brought great comfort and authenticity to the audience.However,most naked-eye 3D displays had some problems,such as single point of view and image crosstalk problems during the viewing process.To solve this problem,by tracking the position of the human eye,the display of the naked eye 3D display was adjusted so as to solve the image crosstalk problem.Therefore,the purpose of this study was to detect and track human eyes under the application background of the tracking slit grating 3D naked eye display.First of all,face detection and tracking methods were innovatively improved.Secondly,an innovative three-layer structure approach was proposed for human eyes detection and tracking.At last,the human eyes to detect and track calculation method was applied to the embedded platform.The main content and results of this article were:1.Explain the display principle of raster 3D naked eye display.Comprehensive analysis of the advantages and disadvantages of domestic and foreign face and human eyes detection and tracking methods.Provide the necessary theoretical basis for the research and research direction of this article.2.A face detection tracking algorithm was proposed based on adaptive boosting(AdaBoost)and spatio-temporal context(STC).First of all,using AdaBoost algorithm to identify the face and extract the face window in the first frame.Then,face tracking was performed by the context of space-time.After that,using histogram of oriented gradient(HOG)algorithm to do similarity judgment.Experiments show that this method achieves about 3 times faster tracking speed than AdaBoost algorithm.3.A tracking positioning algorithm structure was proposed based on AdaBoost,STC-HOG and random forest(RF)On the basis of the face detection and tracking.The algorithm structure was divided into three steps.They were AdaBoost face detection,STC-HOG face tracking and RF human eyes positioning.First of all,face tracking was performed using the content 2 method.Then,the random forest algorithm was used to locate the human eyes.Finally,in order to prove the feasibility of the method,matlab experiments were performed,too.Experimental results indicate that the speed of this method was increased by 2 times in tracking speed,moreover,this method had the same tracking accuracy and robustness as the original algorithm.It can satisfy High precision for human eye location,real-time,good robustness in naked eye 3D display.4.Brief description of the overall architecture of the tracked slit grating 3D nakedeye display system.Combining human eye detection tracking method based on AdaBoost,STC-HOG and RF with human eye detection feedback module.Finally,the GPU-based parallel optimization of the random forest algorithm was performed on the embedded platform.