Research for Free Viewpoint Television System

（1）多通道动态图像分析在我们去年开发的系统中，配备了通信和捕获功能的图像传感器被布置在不同的设置中，形成传感器网络。利用该多通道系统，对捕获的序列进行了空间域和时间域的分析。利用几何变换，我们发现它们有有趣的线和正弦结构。（2）多通道运动图像压缩方法的研究在以往的方法中，编码是基于相邻摄像机之间的相关性。这里，基于上述知识，我们使用时间和空间上的连续图片来执行编码。实验结果表明，该方法可以在较低的码率下获得与原方法相同的图像质量。（3）图像传感器网络的优化为了生成自由视点图像，我们采用了三种方案：1.所有图像在一台计算机上处理; 2.所有图像在多台计算机上共享和处理; 3.部分图像在多台计算机上共享和处理。我们还开发了一种算法，根据系统的条件确定优化方案。因此，我们得到了一个处理时间减少到1/5，与我们以前开发的系统相比。（4）自由视点电视系统的构建通过（1）到（3）的组合，我们实现了真实的时间自由视点电视系统。

(1)Multi-channel Moving Picture AnalysisIn the system we developed last year, the image sensors, equipped with both communication and capturing abilities, were arranged in different setups forming a sensor network. Using this multi-channel system, the captured sequences were analyzed both in Space and Time domains. Using geometric transforms, we found they have interesting line and sinusoidal structures.(2)Research on Compression Method for Multi-channel Moving PicturesWhile in previous methods, coding is based on the correlation between successive cameras. Here based on the knowledge explained above, we performed coding using successive pictures both in time and space. The results showed that we could get the same quality of images as the previous method at a lower bit rate.(3)Optimization of the Image Sensors NetworkIn order to generate free viewpoint images, we proceeded with 3 schemes : 1.All images are processed using one computer 2.All images are shared and processed using multiple computers 3.Images are partially shared and processed using multiple computers. We also developed an algorithm that determines the optimized scheme according to the system's conditions. As a result, we obtained a processing time reduction up to 1/5 in comparison with the system we developed before.(4)Construction of a Free Viewpoint Television SystemAs a result of the combination of (1)to (3), we realized a real time Free Viewpoint Television system.