论文部分内容阅读
文章基于脉动阵列实现HEVC(High Efficiency Video Coding)中8×8的整数DCT(Discrete Cosine Transform)变换,改进通常使用的蝶型算法。整体架构基于脉动阵列的思想,并采用中间值数据重组的设计,使得变换模块可同时实现行列变换操作。只需得到列变换的第一个值便可开始行变换,充分利用了PE单元,减少变换时间并提高计算模块的并行性。文中方法不仅适用于DCT变换,也可用于其他的8×8矩阵相乘,具有通用性。综合结果表明,该设计最高可工作在203.8 MHz的频率上,与其他算法相比时间上只需35个周期,且资源消耗较少。文中方法非常适合于HEVC视频编码对实时性的要求,为HEVC编码标准的硬件实现提供了参考。
In this paper, an 8 × 8 DCT (Discrete Cosine Transform) transform in HEVC (High Efficiency Video Coding) is implemented based on a pulsating array to improve the commonly used butterfly algorithm. The overall architecture is based on the idea of a pulsating array and uses a reorganization of the median data to make the transformation module capable of simultaneously transforming the rows and columns. Just get the first value of the column transform can start the line conversion, take full advantage of the PE unit, reducing the conversion time and improve the parallelism of the computing module. The method in this paper not only applies to DCT transform, but also can be used to multiply other 8 × 8 matrices. The synthesis results show that the design can work at a maximum frequency of 203.8 MHz, compared with other algorithms, it takes only 35 cycles in time and consumes less resources. The method in this paper is very suitable for the real-time requirements of HEVC video coding and provides a reference for the hardware implementation of HEVC coding standard.