在数字信号处理中二进制数量化引起的有限字长效应

The Finite Word Length Effect Caused by Binary Quantification in Digital Signal Processing

在数字信号处理中,采用不同的二进制数的表示方法及量化方式,其误差相距甚远。本文首先分析二进制数的不同表示方式是产生有限字长误差的直接原因,然后对三种二进制数的表示方法及两种量化方式做了详细的说明及分析,同时,还利用MATLAB对不同量化方式的误差敏感性进行了分析,并针对具体问题进行了归纳和总结,提出了相应的解决方案。信号处理在林业森林生态信息处理以及林业生产安全信息处理等方面具有重要的意义。

In processing digital signals, the deviation is far from identification because of the use of different binary representations and quantifications. That the adoption of different binary representations is the immediate cause of finite word length effect is proved in the article first and then detailed explanation and analysis are made of three representations and two quantifications of binary digits. In addition, the deviation＇s sensitivity to different quantifications is analyzed using MATLAB,induction and summary are made about some concrete problems and relevant solutions are presented. Signal processing is of great value in dealing with forestry ecological information and forest safe production information.