Abstract
A new multi-digital-signal-processing (multi-DSP) real-time parallel processing system for image compression is designed to meet the demand of real-time image compression and translation. The system scheme is realized by combining an improved SPIHT algorithm with integer CDF (2, 2) biorthogonal wavelet trasformation, which improves computing speed and makes hardware platform easy. According to the different characteristics of each subimage, the scanning path is changed and the SPIHT algorithm is improved with the parallel processing method of four divided blocks, so that the coding speed is increased and the operation complicacy and time consumption of coding and decoding are reduced. To meet the need of real time, the parallel processing system for image compression is composed of four DSPs with both merits of tight coupled parallel system and loose coupled parallel system. The experiments show that the peak signal to noise ratio (PSNR) of this method based on parallel system raises 10%, and compression time is 30 ms, compared with traditional image compression algorithm.
| Original language | English |
|---|---|
| Pages (from-to) | 1451-1455 |
| Number of pages | 5 |
| Journal | Guangxue Jingmi Gongcheng/Optics and Precision Engineering |
| Volume | 15 |
| Issue number | 9 |
| Publication status | Published - Sept 2007 |
Keywords
- Image compression
- Multi-digital-signal-processing
- Real-time parallel processing system