An Efficient Implementation of Singular Value Decomposition Rotation Transformations with CORDIC Processors
✍ Scribed by X. Hu; S.C. Bass; R.G. Harber
- Publisher
- Elsevier Science
- Year
- 1993
- Tongue
- English
- Weight
- 165 KB
- Volume
- 17
- Category
- Article
- ISSN
- 0743-7315
No coin nor oath required. For personal study only.
✦ Synopsis
One of the most frequently used operations in singular value decomposition (SVD) is the two-sided rotation transformation on (2 \times 2) matrices. In order to achieve high performances for CORDIC SVD processors, a new method of carrying out the two-sided rotation transformation on (2 \times 2) matrices is discussed here. With this new implementation, either the processing time or the hardware cost of SVD processors may be reduced by almost (50 %) when compared to those given by Cavallaro and Luk IProc. SPIE, RealTime Signal Processing IX, San Diego, CA, Aug. 1986, Vol. 698, pp. 45-53]. 1993 Academic Press, Inc.