一种无片外存储的高性能二维DWT架构

发布时间：2018-01-16 10:25

本文关键词：一种无片外存储的高性能二维DWT架构　出处：《西安电子科技大学学报》2017年04期 　论文类型：期刊论文

【摘要】：二维离散小波变换是一种常用的图像处理方法.由于其计算量大,常被设计为硬件电路实现.现有二维离散小波变换的硬件架构均需要存储大量输入图像数据,存储开销较大.为此,笔者设计了一种高性能的二维离散小波变换架构,使用了一种横向扫描的串行二输入无乘法器架构,并消除了片外存储需求.对于一幅N×N的输入图像,系统的总存储需求缩减到10 N字节.另外,文中还使用了正则有符号数乘法器替代传统乘法器,将关键路径延时缩短至约一个加法器延时.经过硬件分析,对比现有其他架构,本架构的总晶体管数量减少4%,硬件效率提高了33%以上.
[Abstract]:Two dimensional discrete wavelet transform is a commonly used method of image processing. Because of its large amount of computation is often designed for hardware implementation. The hardware architecture of the existing two-dimensional discrete wavelet transform to the input image data storage, storage overhead is large. Therefore, the author designed a high performance two-dimensional discrete wavelet transform architecture, the use of the serial input of a horizontal scanning two multiplier architecture, and eliminates the external storage requirements. For the input image of a N * N, the total storage requirement of the system is reduced to 10 N bytes. In addition, the paper also uses the regular number of symbols to replace the traditional multiplier multiplier, will be shortened to critical path delay about an adder delay. After hardware analysis, comparing to other architectures, this architecture reduces the total number of transistor 4%, hardware efficiency is improved by more than 33%.

【作者单位】：天津大学电子信息工程学院;
【基金】：国家自然科学基金资助项目(61474080)
【分类号】：TN47
【正文快照】： 离散小波变换(Discrete Wavelet Transform,DWT)以其优异的时频局部特性和良好的去相关能力,作为多分辨率分析的有效工具,在信号处理和视频压缩等领域获得了广泛的应用[1].由于DWT计算量较大,为了满足实时处理的需求,需要将算法运行在硬件电路上以提升计算速度.因此,设计高效

【相似文献】