摘要为实现基于Kinect的深度图像编码，本文首先研讨了两种主流深度编码方式各自的优劣性能。实验表明，相比于独立深度编码方式，联合深度编码方式在同等失真条件下码率同比降低了约50%，同时其编码时间同比增长了近3倍。其次，基于Kinect的13位原始深度数据，本文主要研究了3种深度图可视化编码方式：高8位数据可视化、低8位数据可视化，以及ROI区域数据集中可视化。实验研究表明，三种可视化方式在不同场景下的适用性有所区别。低8位数据可视化适用于小范围近景；高8位数据可视化适用于大范围远景；而ROI区域数据集中可视化适用于感兴趣区域小范围集中分布的普通场景。此外，本文还探究了三种可视化深度图各自在编码端的表现。深度图场景细节量的增多将复杂化编码端的处理。87882

毕业论文关键词  深度图  深度图编码  HEVC标准  3D-HEVC标准  Kinect 可视化

毕业设计说明书外文摘要

Title  Depth coding based on Kinect         

Abstract To achieve the depth coding based on Kinect, this paper firstly discusses the performance of joint depth coding and independent depth coding。 As shown by the experimental results, the joint depth coding achieves 50% less bit rate reduction compared with the independent depth coding in the same distortion condition, while its encoding time increased by nearly three times。 Secondly, based on the 13-bit original depth data of Kinect, this paper studies three visualization methods for depth map: high 8-bit data visualization, low 8-bit data visualization, and ROI data visualization。 Experimental results show that these visualization methods have different applicability under different scenarios。 Low 8-bit data visualization is suitable to close-range scene。 High 8-bit data visualization is preferred in scenes with long-range depth。 When we are interested in a small range of the common scene, ROI data visualization is more suitable。 Additionally, this paper also explores the performance of different types of depth maps in encoding。 Generally, the increase of details in depth maps would complicate the encoding processing。

Keywords  Depth map, Depth coding, HEVC, 3D-HEVC, Kinect, Visualization

目   次

1 绪论 1

1。1 课题研究背景 1

1。2 研究目的和意义 1

1。3 论文组织结构 2

2  深度编码方式的研讨 3

2。1 HEVC标准 3

2。2 3D-HEVC标准 7

2。3 研讨实验 11

2。4 本章研讨结论 14

3 基于Kinect的深度图编码 16

3。1 Kinect简介 16

3。2 深度测距原理 16

3。3 深度数据流的可视化编码 19

3。4 深度图的预处理 22

3。5 深度图采集实验 23

3。6 深度图编码实验 27

3。7 本章研究结论 28源-于,优Y尔O论U文.网wwW.youeRw.com 原文+QQ75201,8766

结  论