JPEG20000 Algorithm Optimization Based On Altivec Technology

With popularity of the JPEG2000 standard, software solutions applied in different fields have been introduced. Of which JasPer is the most famous one, JasPer is written in C language, with a high platform portability, it has been ported onto Apple, HP, Sun, and many other platforms, compiled and implemented successfully. Because of their belonging to the open source software, users can access source code for free and make reduction and modification on their demand. However, due to the high complexity of JPEG2000 algorithm, the cost on JPEG2000 coding normally is 3~4 times to JPEG does. In JasPer, arithmetic coding and DWT this two modules occupy nearly 80% of the costs as the whole system does. So, to the time-driven applications, they have to choose JPEG, although JPEG2000 is more powerful than JPEG Optimization of JPEG2000 algorithm and application without damage on the attributes is imminent.Much of the research is concerned with the optimization of JPEG2000 algorithm, little on the achievement of Instruction-level optimization study. By studying on multimedia processing instruction sets attached to modern microprocessors, PowerPC microprocessors were chosen as main develop platform in this thsis, which were widely applied in the fieled of communications and image processing. JasPer was optimized with AltiVec technology, AltiVec technology was settled in PowerPC G4 series microprocessors by Motorola, it is one subset of SIMD instruction sets, the bottleneck of process can be easily sovled by using it. Experimental result indicates that the improvation is effective.