Research On 3D Integral Imaging Visualization Based On Depth Information

3D imaging and display technologies have become a hot research topic in these years. The use of micro lens array in traditional integral imaging system will cause the low resolution of the element images and unclear effect of display. If some regions of the target object are occluded by the occlusions when we pick up the elemental images, the focused plane images of the target objects will be covered with some defocused occlusion information when we reconstruct the captured elemental images computationally. In order to overcome the problem of low resolution of the reconstructed object plane images caused by occlusion, this paper proposes the technologies of occlusion effect removal in integral imaging system. Besides, in order to implement optical display of the off-axially distributed based integral imaging system, an optical display method based on off-axially depth extraction is proposed in this paper.In the second chapter of this paper, a mean shift segmentation based occlusion removal algorithm is presented in the traditional 3D integrated imaging. The main procedures are as follows:Firstly, capture the elemental image array by means of pickup technology in 3D integral imaging, and transform the elemental image array into sub-image array; Secondly, implement the mean shift segmentation based stereo matching algorithm for the adjacent sub-images in the sub-image array, and then get the disparity maps of the sub-images. Mark the occlusion pixels in the sub-image based on the obtained disparity map; Then, restore the missing object information in the sub-images by using of the stereo matching based pixel filling techniques; Finally, transform the modified sub-image array into elemental image array and reconstruct the target object computationally, an information completed target plane image without occlusion effect will be obtained.In the third chapter of this paper, a mean shift segmentation based stereo matching algorithm and a sparse representation based image inpainting algorithm are used in the occlusion removal of three-dimensional synthetic aperture integral imaging. The main steps of this technology are as follows:Firstly, capture the elemental images by means of pick up technology in 3D synthetic aperture integral imaging; Secondly, combine the block matching algorithm with mean shift segmentation algorithm to obtain the disparity maps of the elemental images. Mark the occlusion positions in the elemental image based on the obtained disparity map; Then, restore the missing target information of the elemental image using sparsity represented patch propagation based elemental image inpainting algorithm; Finally, the target information completed plane image without occlusion effect will be obtained with the computational reconstruction algorithm.In the fourth chapter of this paper, a 3D integral imaging system which gets the elemental images with off-axially distributed image sensing is proposed.The steps of display of elemental images picked up in off-axially distributed image sensing system are as follows: Firstly, pick up the elemental images by using off-axially distributed image sensing technique; Secondly, extract the depth map and synthesize the color image corresponding to the first picked up elemental image using profilometry. Then synthesize the elemental image array based on the optic geometry of the 3D integral imaging. Finally, we can obtain a higher resolution of the reconstructed image by computational reconstruction and display the 3D object in free space by optical reconstruction in 3D integral imaging.The experimental results show that the proposed occlusion removal techniques in integral imaging system can effectively improve the clarity of the target object. An off-axially distributed image sensing based 3D integral imaging display system can overcome the drawback that the resolution of the picked up elemental image array in traditional integral imaging system is low, and it can also solve the problem that the distance between 3D objects and the camera is limited when we pick up the elemental images in traditional integral imaging system in a certain extent.