Research On Integral Imaging 3D Display Techniques

Integral imaging has attracted attention as a new type of true 3D display technique because of its excellent display quality,equipment-free characteristic,and lack of fatigue.Studies on integral imaging have focused on the acquisition of the elementary image array,reconstruction of displayed images,and applications in display devices.Integral imaging can be of three main types: total optical integral imaging,computer-assisted integral imaging,and virtual reconstruction integral imaging.In virtual reconstruction integral imaging,we use an optical lens array and devices such as a charge-coupled device(CCD)to pick up an elementary image array,and then,we reconstruct virtual images based on this array.Through this procedure,we can obtain images of different depths or from different viewing points.However,there are still shortcomings in integral imaging based on computer reconstruction.First of all,the elementary image array for virtual viewing point reconstruction is picked up by real photoelectric device,crosstalk and overlapping of light will inevitably occur in the picking-up process which will cause deviations in elementary image;secondly,in the process of virtual image reconstruction,if we use the traditional non-periodic pixel extraction algorithm,we discard the extracted pixels when the pixels extracted are beyond the range of matching elementary image which will narrow the range of the reconstruction image and lower the utilization efficiency of the elementary image array.In order to eliminate the defects in both picking-up and reconstruction process,we propose the integral imaging reconstruction algorithm based on virtual picking-up system which is capable to solve the problems in both aspects.First of all,in the process of picking-up scenes and generating elementary image array,we use the principle in computer-assisted integral imaging for reference and transform the picking-up system from real optical system to virtual picking-up system.We use software called maya to build up virtual picking-up system and simulate real cameraswith virtual cameras in maya by setting the parameters of virtual cameras like focal length and apertures.We use the script within maya called mel to generate picking-up camera array automatically and create virtual scene for shooting.Every elementary image is specifically generated by the matching camera with no redundant light picked up,then we can generate the elementary image array by joining every elementary image which successfully avoid the interrupt of redundant light.This is vital for further experiment,only we preclude the interrupt of the redundant light can we confirm that the distortion of reconstruction image is derived from mismatching of the extracted pixels.In the process of image reconstruction,we improve the tradition non-periodic pixel extraction algorithm based on virtual picking-up system.When the extracted pixel is beyond the range of its matching elementary image,we search for the matching pixel of that pixel in other elementary images using light-ray tracing principle instead of discarding the pixel.First we analyze the principle of distortion and method of searching for matching pixels,then we merge the pixels with same substantial into same distortion block and getting different matching blocks of different distortion blocks using different tracing light route,finally we replace the distortion block with the matching block which indicates that the range of the reconstruction image is broadened and the utilization of the elementary image array is improved.We also compare the reconstruction generated by these two algorithm and calculate the improvement ratio.The result shows that the virtual picking-up system and improved reconstruction algorithm that we proposed perform excellently,it offers ideal experimental environment and expands the showing range of the reconstruction image,it is of great significance in both theoretical research and practical reconstruction.