| In order to enhance the immersion of virtual reality,the visual positioning technology based on markers achieved displacement synchronization between real space and virtual space by tracking the user’s head position,so that users have a sense of presence.However,this positioning system was limited by the user action and the indoor environment,such as the user moves too fast,the indoor light variance in image would lead to failure to locate and then cause the displacement synchronization to fail and affect the user experience.Based on researching the virtual reality visual positioning technology,finding and analyzing the profound reasons cause it to fail,the work of this thesis aims to research the fault-tolerant algorithm for vision positioning technology and the image processing algorithms in the uneven illumination environment,so as to improve the fault tolerance and robustness of the system,as given in the following.Firstly,a fault-tolerant algorithm based on inertial positioning technology is proposed to solve the problem of user moving too fast.The method uses head mounted display with the inertial instrument to obtain user head motion vector information,and it uses the visual positioning data as the initial data and then uses the kalman filter to calculate the acceleration integral,so as to complete the position estimation in the case of visual positioning failure.Experimental results show that the proposed method can complement the lack of positioning data,improve the system fault tolerance and robustness.Secondly,aiming at the problem of visual localization failure caused by uneven indoor illumination,an adaptive visual tracking algorithm based on illumination detection is proposed.Firstly,the image illumination environment is detected by the gray value distribution.When the detected image might be in uneven illumination environment,it uses OpenCV local adaptive threshold method for image segmentation,or else when the detection result is uniform illumination,it uses a fixed threshold segmentation method for image processing.Then the alrorithm completes the visual tracking by means of marker tracking based on ARToolkit.Experimental results show that the proposed algorithm has strong adaptability in the environment of uneven illumination,and it improves the robustness and ensures the efficiency of the algorithm.Thirdly,an immersive virtual reality system is designed and implemented based on the above researches.It allows the user to move in virtual space by walking in the real world,and we can observe the virtual world position changes through the Oculus Rift DK2 head mounted display. |
