Image data recording system on the Multi-media database

最近，个人电脑和多媒体系统在日常生活中越来越受欢迎。因此，有一些机会来处理数字化图像，特别是运动图像。由于图像数据量大，需要对图像数据进行编码进行传输和/或存储。因此，已经建立了一些数字图像编码的国际标准。然而，人们希望计算机处理的图像变得多种多样，这就需要更加通用的编码算法。在本研究中，我们讨论了提高压缩性能的编码方案和现有标准的高速实现。为了实现更高的编码效率，我们首先关注超高清图像的无损编码，然后开发出优于国际标准JPEG的编码方案。接下来，我们提出了将Golomb编码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（标准差）来简化Intra/Non-Intra决策过程的方法。此外，我们还提出了一种利用已解码的运动图像序列插值高分辨率静止图像的方案。该方法非常简单，只需要运动矢量，而运动矢量是MPEG参数所涉及的。

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.