Application of COPERM to Image and EEG signal Compression

• Main Authors: Ying-Chan Huang, 黃英展
• Other Authors: Pei-Chen Lo
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/91109385216265857811

碩士 === 國立交通大學 === 電機與控制工程系 === 88 === Abstract： A number of methods have been developed to compress both the image data and time sequences. Each of them was designed to simultaneously optimize the compression ratio, computational efficiency, and the quality of decompressed image. This thesis discusses a method of data compression, transform-domain energy Compaction by Optimal PERMutation (COPERM), which is particular- ly effective in compressing the wide-band data.The performance of COPERM is affected by some factors, for example, the number of re-quantized levels and the number of frequency components to be coded. Hence, we will first investigate the effect of these factors on the compression ratio and quality of the decompressed image. The results provide a guideline to determine the implementing parameters. We then apply the COPERM to the EEG (electroencephalograph) signal compression. The resulting compression ratio is lower than the direct DCT (discrete cosine transform) method without permutation. Further study shows that the EEG spectrum is considered to be narrow-band for the COPERM. When applying to the wide-band, high-frequency signals like the speech and EMG (electromyogram) signals, the COPERM proves effective.