权益分类	功能权益	普通用户	{{item.name}}会员
{{category.name}}	{{benefitItem.name}}

Image data recording system on the Multi-media database

多媒体数据库图像数据记录系统

基本信息

批准号：
09680410
负责人：
KUROKI Yoshimitsu
金额：
$ 0.26万
依托单位：
Kagoshima National College of Technology
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
1997
资助国家：
日本
起止时间：
1997 至 1998
项目状态：
已结题

来源：
https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-09680410/
关键词：
image processing image coding MPEG JPEG motion vector Golomb code super high definition image 動画像符号化動き補償予測マクロブロック標準偏差ラプラス分布

项目摘要

Recently, personal computers and Multi-media systems become much more popular in the daily life. Accordingly, there are some opportunities to deal with digitized images, especially moving images. Due to the large amount of the image data, it is necessary to encode the image data to transmit and/or store. Some international standards of digital image coding, therefore, have been established. However, the images one would like to handle by the computer become variety which requires much more universal coding algorithms. In this study, we discussed coding schemes to improve the compression performance and a high-speed implementation of the existing standards. To achieve higher coding efficiency, first, we focused the lossless coding of super high definition image, and then exploited the coding scheme superior to the international standard JPEG.Next, we presented that adaptation of the Golomb code to the coding of predictive motion vector residual decreases the quantity of the coding data size compared with the MPEG strategy. Then, we described the method to simplify the Intra/Non-Intra decision procedure by estimate the variances (standard deviations) of motion compensated predictive residuals. Moreover, we proposed a scheme to interpolate high-resolution still images by utilizing the already-decoded moving image sequence. The proposed method is very simple because only motion vectors, which involved in the MPEG parameters, are required.

近年来，个人电脑和多媒体系统在日常生活中越来越普及。因此，存在一些处理数字化图像，特别是运动图像的机会。由于图像数据量大，因此需要对图像数据进行编码以进行发送和/或存储。因此，一些数字图像编码的国际标准已经建立。然而，人们希望通过计算机处理的图像变得多样化，这需要更通用的编码算法。在这项研究中，我们讨论了编码方案，以提高压缩性能和高速实现现有的标准。为了提高编码效率，本文首先研究了超高清图像的无损编码，然后研究了优于国际标准JPEG的上级编码方案，并提出了将Golomb码应用于预测运动矢量残差的编码，与MPEG策略相比，减少了编码数据量。然后，我们描述了通过估计运动补偿预测残差的方差（标准偏差）来简化帧内/非帧内判决过程的方法。此外，我们提出了一种方案，通过利用已解码的运动图像序列插入高分辨率的静止图像。该方法只需要运动矢量，而运动矢量是MPEG参数中的一部分。