摘要
随着数字信号处理在计算机领域的广泛应用,各类正交变换的应用和研究也日益深入,尤其在图像处理和语音识别方面。其中应用最广泛的变换就是离散余弦变换(dct)及在其基础上发展的各类快速算法,如pruning fast dct(简写为pruning fct)等。本文介绍了一种运行于dsp处理器的pruning fct改进算法,在该算法中,通过对pruning fct系数的分解,减少了pruning dct系数的个数,合并了算法中的步骤,由此减少了运算中内存访问的次数,在TI TMSC320C64x DSP处理器上的运行结果显示,与未修改的pruning fct算法相比,改进后的算法可以减少大约40%的内存访问次数和30%的内存空间。
With the widely use of the digital signal processing, various applications and research of orthogonal transtormation develop last. Especially in the fields of image processing and speech recognition. The most widely used method is Discrete Cosine Transform （DCT） and other algorithms based on it, such as Pruning Fast DCT. This paper proposed an improved algorithm running on the DSP platform. This algot＇ithm reduce the number of weighting factors by decomposing the weighting factors at first.Then it combines the butterflies slages：Fhese operations can effcclively reduce the memory accesses.Experiment running on DSP platform shows that the proposed method can achieve average of 40% memory access redm：tioo and 30% memory space saving.
出处
《科技信息》
2011年第13期461-463,共3页
Science & Technology Information
