| We have used the single camera to capture the visual information of the world for along time, but this method can not bring the sense of depth, the stereo feeling and thecomprehensive understanding of the object to human eyes. With the development of therelated disciplines and the increased demand for the visual effect, three dimensionaldisplay technology appeared and quickly became the new display mode, which has beenstudied by all the countries around the world. In recent years, three dimensional displaytechnology has been widely applied to entertainment, industrial manufacture, medicaltreatment and public health, education and training, military and so on.At present, the existing three dimensional display technology mainly includes theparallax-based three dimensional display technology and the true three dimensionaldisplay technology. The parallax-based three dimensional display technology violates thenormal working mechanism of human eyes in the real world, and it can cause the visualfatigue if people watch the display screen for a long time, so it is not the ideal method torealize three dimensional display technology. The true three dimensional displaytechnology realizes the reconstruction of the stereo image in the space by adding the lightstogether, and the viewers can perceive the depth information of the stereo image byphysiologically regulating the focal length of the eyes. So the true three dimensionaldisplay technology can not cause the visual fatigue, and is widely considered as the trendof three dimensional display technology. The true three dimensional display technologymainly includes integral imaging, holography and volumetric display. Compared to theholography and volumetric display, whose temporal-spatial resolutions are limited, theintegral imaging not only allows display of full color three dimensional images withcontinuous parallax within a wide viewing zone, but also has a strong feasibility and widepromotion space owing to the less limitation on the display device.Integral imaging mainly faces the following challenging problems:(1) Generate the elemental image array of the true three dimensional scenesDirect acquisition of true three dimensional scenes can avoid the complex modelingsteps in the process of computer synthesis and enrich the display material of integralimaging, however, because there are a large number of viewpoint images in the elementalimage array, the current equipment and the traditional method both encounter the problems of low resolution, complicated hardware structure, difficulties in cameraadjustment, high cost and so on.(2) Efficient compression coding of elemental image arrayThe same as all other stereo display technologies, integral imaging is also faced withthe problem of how to store and transmit the large amounts of data flow. In integralimaging, an elemental image array consists of thousands of small-size elemental images,and each elemental image corresponds to one viewpoint image. Most existing stereoimage coding methods are not suitable for the elemental image array since they aim atcompression of several large-size viewpoint images.(3) Reconstruction of the stereo image in the large screenReconstruction of the stereo image in the large screen can improve the effect of depthinformation and bring the viewers great visual impact and infectivity. Currently, the mainproblems include the high cost of the lenslet array and the immature manufacturingprocess, so the key to the implementation of the technology lies in the improvement of themanufacturing process of lenslet array or adopting other optics devices, whose opticalperformance are the same as the lenslet array, according to the optics principle of integralimaging.According to the above problems, this paper proposed the following five methods,which realize real scene pickup, virtual viewpoint synthesis, compression coding anddisplay, respectively. These methods, to some extent, break through some current technicalbottlenecks and lay a theoretical foundation for the practical application and the furtherdevelopment of integral imaging:(1) Elemental image array generation based on sparse viewpoint pickupAfter further study on the lenslet-based pickup scheme, we put forward this methodwhich combines the direct acquisition with the computer synthesis. First, take advantageof the sparse viewpoint images which are captured by the camera array to reconstruct thespatial location and the structure information of the objects. Then, generate the elementalimage array by parallel mapping the reconstructed object into every virtual lens.Compared with the lenslet-based pickup scheme, the proposed method can generatehigh-resolution elemental image arrays since it is not limited by the pickup device, and itcan significantly decrease the costs and the workload by comparison to the traditionalcamera array pickup method.(2) Elemental image array generation based on discrete viewpoint pickup andwindow interceptionThe theoretical foundation of this method is the optical mapping relationship between elemental image array and sub-image array. Considering the fact that the number of thesub-images in the sub-image array is far less than the number of the elemental images inthe elemental image array, we propose a converse approach by which the elemental imagearray can be obtained from the sub-image array of the object which has already beenobtained in advance. Compared with the elemental image array generation method basedon sparse viewpoint pickup which is mentioned, the elemental image array generated bythis method has better view and display depth.(3) Two dimensional virtual viewpoint synthesis based on the depth image basedrendering (DIBR) technologyThis method extends the one dimensional virtual viewpoint synthesis method to twodimensional by adding the two-step mapping algorithm. And this method can not onlyimprove the pickup efficiency of the elemental image array in integral imaging but alsomake the generated elemental image array adapt to the needs of different displayplatforms.(4) Distributed prediction coding for elemental image arrayIn this method, the distributed prediction coding framework is used to compress thesub-image array which is obtained from the elemental image array. According to theunique correlation distribution of the sub-image array, we design a simple and effectiveprediction model. This method is especially suitable for the mobile devices whosecomputation and energy consumption are limited since the distributed coding method cantransfer the computation complexity from the encoder to the decoder.(5) Integral imaging display platform based on lenslet arrayIn this method, we analyze systematically the display characteristics of integralimaging and build the integral imaging display platforms by method of photoelectriccombination, which includes the high resolution projector combined with lenslet array andthe flat-panel LCD display combined with lenticular lens array. Meanwhile, according tothe optical properties of lenticular lens array, a novel method is proposed forreconstructing the viewpoint image by computer during the one dimensional integralimaging display process, and this method can be used as a simulation analysis method ofone dimensional integral imaging display. |
PDF Full Text Request