Improved digital filter structure for fast moments computation

Abstract

An efficient approach is introduced to implement a digital filter structure for fast moments calculation. The approach not only improves the computation efficiency, but also simplifies possible realisations using both hardware and software. A suitable filter structure for the computation of moments is reviewed. The improved filter structure is compared with conventional algorithms, and yields superior performance over other algorithms in both computational speed and complexity. The most time-consuming multiplication operations in traditional approaches can be completely eliminated, and over 60% of additions in the Straightforward method can be avoided. Experimental results show that the approach requires only 20% and 88% of the computation times of the Straightforward and Hatamian’s digital filter approaches, respectively.