Study on a Coding and Representation Method of Super High Definition Images for Digital Museum

重要的是要把伟大的艺术作品和史料留给后人。数字存档是保存这些材料的一种方式，不会有老化、变质或损坏的风险。如果将这些数字档案存储在数据库中，每个数据库通过网络连接起来，我们也可以不受空间限制地研究或享受这些材料的信息。为此，许多博物馆和研究机构建立了各种虚拟博物馆。为了高保真地呈现艺术作品，这些材料应该被数字化为超高清图像。此外，对于立体材料，如雕像或陶器，要求从多个视角拍摄照片。因此，需要对这些海量的数字数据进行压缩存储和传输。然而，对于这些图像数据的编码，诸如JPEG或MPEG的有损压缩方案是不可取的，因为它们会导致图像降级。为了表现艺术作品的细节，需要引入无损压缩方案来保存艺术作品的数字图像信息。从这一角度出发，我们提出了一种超高清图像的无损编码方案。在这项研究中，我们提出了一种新的从不同角度获取的凸物体数字图像序列的无损编码方案。在文献中已经提出了各种多视点图像的编码方案。这些算法大多是利用相邻帧或三维几何模型的信息进行视差估计和补偿。该方法对于多视点图像的编码是有效的。然而，这些研究都是针对有损压缩性能进行的。我们的目标是对海量的多视点图像数据进行无损编码。

It is important to leave great art works and historical materials to the future generations. Digital archiving is a way of preserving these materials without risks of aged deterioration or damage. If these digital archives are stored in databases and each database is connected by network, we will also be able to study or enjoy the information of these materials without the spatial restrictions. For this reason, various virtual museums have been built by many museums and research organizations.In order to represent art works with high fidelity, these materials should be digitized as super high definition images. Moreover, for tri-dimensional materials, such as statues or potteries, it is required to take photographs from multiple view points. Therefore, huge amount of these digital data should be compressed for storage and transmission. However, lossy compression schemes such as JPEG or MPEG are not desirable for coding of these image data, because they cause image degradation. In order to represent details of art works, lossless compression schemes should be introduced to preserve digital image information of the art works. We have been proposed a lossless coding scheme for super high definition images from above standpoint.In this study, we propose a new lossless coding scheme for digital image sequences of convex objects which were taken from different perspectives. Various coding schemes of multi-viewpoint images have been proposed in the literature. Most of these schemes are based on the disparity estimation and compensation using the information given by neighboring frames or 3-D geometry model. This approach is effective for coding of multi-viewpoint image. However, these researches studied on lossy compression performance. We aim to encode huge amount of multi-viewpoint image data in a lossless manner.