數位類比混合系統應用於圖片傳輸

Abstract

在本篇論文裡，我們提出一種將數位和類比結合的混合系統，此系統在稍後的模擬結果裡證實可以有效的提升JPEG的峰值訊雜比，數位類比混合系統的優勢在於它在高通道訊雜比及低通道訊雜比的表現都很不錯，這些要歸功於它同時有數位及類比系統的關係。但也因為有類比系統的關係，所以存在著跳線問題，也因為有跳線問題所以空間填充曲線的分形維度及縮放比例要謹慎挑選。數位類比混合系統比起傳統純數位系統有個很大的優勢那就是純數位系統的量化誤差是不可能再被復原的，而數位類比混合系統既可以維持純數位系統的優勢又可以將量化誤差透過類比分支傳送到接收端並在接收端將資料補正，也因為這樣所以我們選擇JPEG壓縮來當作我們系統的應用。在JPEG壓縮的流程中會將部分高頻的成分透過量化的方式去除掉，這時若是用我們的系統來傳輸資料的話我們就能將那些被去除掉的高頻成分透過類比分支傳輸過去，這樣的做法既可以增加圖片的峰值訊雜比又不用增加或增加傳輸能量。然而在本篇論文裡有做一個假設那就是我們並沒有在接收端做JPEG圖片的錯誤更正，也沒有在傳送的過程中做資料的保護，換句話說本篇都假設在數位類比混合系統裡的數位分支在接收端是沒有誤差的。

In this thesis, we propose Hybrid Digital-Analog (HDA) system that is able to improve PSNR of JPEG image. HDA system has good performance at both high CSNR and low CSNR, because it has digital branch and analog branch simultaneously. However, when we use the HDA system, we should be aware of the jump line issue in the analog branch. Parameters of the space-filling curve such as scale and fractal dimension should be chosen sophisticatedly. HDA system usually can solve the problem of the digital system where quantization error cannot be recovered at the receiver side. This suggests us to choose JPEG compression to implement HDA. JPEG compression will diminish the high-frequency component in quantization step. If transmit JPEG image by HDA system, system can transmit quantization error. Those quantization errors are some components which are discarded by compression procedure. Transmitting JPEG image with this method would improve PSNR without increasing the amount of channel using or channel power. In this thesis, we do not propose the error recovery or protected method of JPEG image. That is a future work. Because of the error recovery problem, we set the channel CSNR (channel SNR) to be high so that we canmake sure error free of digital branch at receiver.