Fractal audio compression.

number: 
1087
إنجليزية
department: 
Degree: 
Imprint: 
Physics
Author: 
Wesam fawzi jassim Mohammed
Supervisor: 
Dr.Loay A. George
Dr. Laith A. Al-Ani
year: 
2005
Abstract:

Fractal audio compression is based on the concept of a partitioned iterated function system (PIFS). Fractal audio compression exploits the affine redundancy that is commonly present in audio; this redundancy is related to the similarity of an audio with itself. In other words, fractal audio compression finds similar patterns that exist in different scales and different places in audio, and then eliminates as much redundancy as possible. In this work the possibility of implementing fractal audio compression is investigated. The implemented system consists of two major units; the first is the Encoding unit and the second is the Decoding unit. Encoding is done by partitioning the range pool (which is the original audio) into non-overlapping blocks, called range blocks, and partitioning the domain pool (which is the result of the original audio after down sampling) into overlapped blocks with the same size of range blocks called domain blocks. A fixed size-partitioning scheme is used to partition the domain pool and the range pool. After generating the range and domain pools, for every range block, the best-matching domain block in the domain pool is searched for by performing a set of affine transformations on them. Thus the encoding is completed by saving the optimal affine parameters for every range block. The Decoding process can be done by repeatedly applying the affine transformation on an initially blank audio and its subsequent reconstructed audio, until it completely reconstruct an approximate wave to the original audio. The time required to compress an audio file is affected by the size of each block being extracted from the proceed audio file; this means smaller block size implies longer time required to compress the corresponding audio file. The implemented system was tested using five wave samples of data. The proposed work was implemented by using Visual Basic (6.0) as a programming language, the fidelity measure MSE and PSNR were used to check the results of the whole implemented technique. The best results obtained from the implemented system were for the test (sample -1) in case were block size equal to (60) the compression ratio is (15.51:1), also for test (sample-2) were block size equal to (40) the compression ratio is (11.03:1), and for small block size as in test (sample-3) were the block size is (20) the value of the PSNR is good that its equal to (35.76 dB), also for test sample-1 when the block size is (10) the PSNR value (45.63 dB).