Efficient Scalable Analysis and Coding of Hypervolume Data

点击翻译按钮获取中文摘要

Volumetric data records with three and more dimensions appear in many sectors of natural and engineering science. Dynamic 3-D volumes from magnetic resonance tomography (MRT) and computed tomography (CT) become more and more important in medical image processing. So far, this project focused on improving the analysis of high dimensional (hyper-) volume data by using compensated wavelet lifting. The goal is to obtain an improved scalable representation and feasible compensation methods were developed therefore. By their incorporation directly into the lifting structure, the structures and characteristics of the wavelet coefficients are modified fundamentally so existing methods for coding cannot operate in an optimum way anymore.In this application, novel more efficient methods for scalable coding are developed. Therefor, graph-based approaches are used. The usage of one specific coder is set aside to obtain a higher coding efficiency. Within the scope of the present research of the applicant and the literature, considerable coding gains were achieved by combining hybrid coding and specialized residual coding methods for obtaining scalable lossless coding of video data. Thus, the combination of wavelet-based and hybrid approaches is addressed to enable a more flexible Decomposition as well as a higher coding efficiency for (hyper-) volume data.