Pengkodean Gabungan Sumber-Kanal Dengan Kompresi Fraktal Untuk Transmisi Citra: Joint Source-Channel Fractal Decoding For Image Transmission

Rosdiana, Rosdiana (2005) Pengkodean Gabungan Sumber-Kanal Dengan Kompresi Fraktal Untuk Transmisi Citra: Joint Source-Channel Fractal Decoding For Image Transmission. Masters thesis, Institut Teknologi Sepuluh November.

Text 2202203004-Master_thesis.pdf Restricted to Repository staff only Download (28MB)

Abstract

Kompresi fraktal sangat sensitif terhadap kesalahan bit kanal, karena proses pendekodean parameter fraktal tidak hanya bergantung pada blok partisinya sendiri namun bergantung pula pada blok lainnya. Untuk itu dirancang sistem pengkodean gabungan sumber-kanal, yang dapat meminimalkan kesalahan tersebut dengan cara memberikan umpan balik dari sumber ke kanal atau sebaliknya. Metode penelitian yang digunakan adalah mengirim bit-bit dengan proteksi EEP (Equal Error Protection), memberikan umpan balik EFSIF (Error Free Source Information Feedback) deteksi/koreksi kesalahan Fraktal menuju pengkode Turbo dan memodifikasi bit-bit informasi ekstrinsik di pengkode Turbo dari nilai-nilai CSV (Channel Soft Value)nya. Unjuk kerja sistem dilihat dari sisi kuantitas (Rate-Distortion dan PSNR-BER) dan kualitas (fidelity visualisasi) nya. Sebagai referensi, kompresi fraktal dibandingkan dengan JPEG dan kanal Rayleigh dibandingkan dengan kanal AWGN. Hasil penelitian menunjukkan bahwa untuk rasio kompresi 95.31 % dan partisi 8x8 bit/blok, distorsi fraktal (D=539.95) lebih buruk daripada distorsi JPEG (D=334.12). Namun visualisasi fraktal lebih baik daripada JPEG. Karena pada fraktal tidak terdapat artifact blocking sebagaimana JPEG karena ada self similarity. Semakin kecil ukuran partisi blok, unjuk kerja fraktal semakin meningkat. Berkebalikan. Adapun hasil penelitian joint coding dengan memodifikasi bit-bit ekstrinsik-EEP, tidak menunjukkan perbedaan yang nyata dengan tandem coding. Hal ini disebabkan karena kinerja turbo sangat dipengaruhi oleh ukuran frame dan laju bit. Sedangkan pengaruh jumlah/pola generator, jumlah iterasi di pendekode dan jumlah kesalahan dalam frame tidak begitu berpengaruh banyak.
=====================================================================================================================================
Fractal image codes are very sensitive to the channel errors because the decoding of a block is dependent not only on the code information associated to this block but also on the code information associated to other blocks. Therefore, we proposed an error control coding on the Joint Source-Channel Decoding to minimize errors with feedbacks. Turbo is such as a channel code and Rayleigh fading is such as a channel model. The experiment methods are sending bits with EEP (Equal Error Protection), giving feedback from the source error & detected corrector to the MAP Turbo decoder and modifying EFSIF (Error Free Source Information Feedback) in Turbo decoder from it's CSV (Channel Soft Value). The measurements of perfomancy system are quantity measurements (Rate-Distortion and PSNR_BER) and quality measurement (fidelity of visualisation). We used lena.tif gray scale 256x256 as an experiment image. We compared fractal image compression with JPEG as a reference and also Rayleigh fading channel with AWGN channel. The results show that for the same 95.31% compression ratio and 8x8 bit block partition, the quantities (Rate, Distortion, PSNR) of fractal image compression are worse than JPEG. But in a point of viewing, fractal is better than JPEG. There are no artifacts blockings in fractal's, like in JPEG's because of fractal's self similarity. As for joint coding based on an EEP - a modification of the extrinsict information passed between turbo decoder, there are no significant perfomancy improvements. It is rather the same as like tandem coding. Because the exact improvement comes from the increasing frame size and decreasing bit rate and not from the increasing of encoder memori size, the increasing of number decoder iteration or the decreasing of frame error.

Item Type:	Thesis (Masters)
Additional Information:	RTE 006.42 Ros p
Uncontrolled Keywords:	Fraktal, Turbo, Joint Source-Channel Coding, Quadtree Partition, Equal Error Protection, modifikasi bit-bit ekstrinsik
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions:	Faculty of Electrical Technology > Electrical Engineering > 20101-(S2) Master Thesis
Depositing User:	magang .
Date Deposited:	06 Oct 2026 08:01
Last Modified:	06 Oct 2026 08:01
URI:	http://repository.its.ac.id/id/eprint/145344

Actions (login required)

View Item