Theoretical Analysis And Experimental Research On Still Multi-View Auto-Stereo Image Processing

Stereo principle, display, synthesis and compression are the four key factors of stereo image. This paper has done theoretical analysis and experimental research on the above factors, including visual principle characteristic of still stereo image, related image processing for multi-view still auto-stereo image composition display, synthesizing for multi-view still auto-stereo image, and compression for multi-view still auto-stereo image.Firstly, based on HVS characteristic, the affect of luminance, contrast, resolution and chroma to two-view stereo imaging system has been studied. An important result is found for the first time: chroma difference between different views almost has no effect on stereo imaging. With chroma substitution, the data amount of two-view stereo image has been effectively reduced. Furthermore, with chroma substitution, the data amount of multi-view stereo image has been reduced more.Secondly, to solve spectrum aliasing in the multi-view auto-stereo image multiplexing process, pre-filtering prior to multiplexing is studied based on the theory of two-dimensional sampling. The corresponding theoretical sub-sampling model is proposed. The two-dimensional anti-alias digital filter is designed. The designed two-dimensional anti-alias digital filter is applied to multi-view auto-stereo image processing and it greatly removes the distortion caused by spectrum aliasing. The experimental result fits the theoretical analysis very well.Thirdly, to improve the synthesizing speed and reduce the complexity of the current stereo synthesizing algorithms, three kinds of high efficient stereo image synthesizing algorithms are proposed, which include AIV, grads and SSPL synthesizing algorithm. The proposed algorithms can be easily applied to real-time processing. They can synthesize multi-view stereo image quickly and well.Lastly, breaking through the limitation of 2D image compression schemes, a still multi-view auto-stereo image compression scheme based on H.264 is proposed. The stereo image quality and compression efficiency of this scheme is much higher than the classical JEPG scheme. Then, a novel scalable compression scheme for still multi-view auto-stereo image is proposed, which is very suitable for the transport bandwidth fluctuation. Moreover, image quality preference coding scheme and stereo effect preference coding scheme are designed for different application area. What's more, another scalable compression scheme based on chroma substitution for still multi-view auto-stereo image is proposed for the first time, which can improve the compression efficiency more.