基于GPU的应用于数字电影图像压缩算法中的EBCOT位平面编码
首发时间:2011-11-14
摘要:相比目前主流的图片格式JPEG来说,JPEG2000在压缩效率和码流可伸缩性方面都有明显的优势,被业界公认为下一代静态图像压缩标准,数字影院倡导组织(DCI,Digital Cinema Initiatives)已将其作为数字电影的图像格式。但是由于其编码过程,特别是EBCOT(Embedded Block Coding with Optimized Truncation)部分,算法复杂度较大,制约了它在实时性要求较高的系统中的应用。随着图形处理器(GPU)的发展,使用GPU辅助CPU进行大数据量运算已经成为可能。NVIDIA公司2006年正式发布计算统一设备架构(CUDA),采用SIMT的模式,使GPU用于通用计算更加容易和有效。在本文中,我们设计了一种新的基于CUDA的并行EBCOT位平面编码算法,测试结果显示,相比传统的CPU串行编码,有3-4倍的加速比。
For information in English, please click here
GPU-Based Bit- Plane Coding for EBCOT in Image Compression Algorithm of Digital Cinema
Abstract:JPEG2000 is the future standard of static compressed image. It combines better compression performance and code stream with good scalability compared with JPEG. Digital Cinema Initiatives(DCI) has selected it as the compression format to be used for digital distribution of motion pictures. However, due to its complexity of algorithm especially the Embedded Block Coding with Optimized Truncation(EBCOT) and time-consuming, it hasn't been widely used in systems with high real-time requirement.As the rapidly development of Graphics Processing Unit(GPU), it is possible to compute massive data using GPU other than CPU. Especially, Compute Unified Device Architecture(CUDA) proposed by NVIDIA in 2006 makes it easier and more effective to process data using Single Instruction Multiple Threads(SIMT) pattern.This paper proposes a novel GPU-Based Bit-Plane Coding(BPC) for EBCOT in Image Compression Algorithm of Digital Cinema. Our experimental result demonstrates that it is 3 to 4 times speedup compared with the original implementation on the CPU.
Keywords: image process digital film EBCOT GPU CUDA
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
基于GPU的应用于数字电影图像压缩算法中的EBCOT位平面编码
评论
全部评论0/1000