ADWA: A filtering paradigm for signal's noise removal and feature preservation

Abstract While denoising a signal, the existing filtering methods are beset with the contradiction between signal’s noise removal and feature preservation; in face of this denoising contradiction, there is no effective solution by now and they have to make a trade-off between these two contradictory aspects. With the purpose to further alleviate this contradiction progressively, but not to make a trade-off, here we established a universal denoising paradigm named as attribute distance weighted average (ADWA) from the perspective of signal’s attribute analysis. ADWA not only breaks the existing methods’ limitations on the number and kinds of signal’s attributes, and has a flexible expandability to include any attributes, but also includes the ideas and takes the advantages of many existing filtering methods. Both theoretical analysis and experimental results show that, for any noisy signal, under the condition of a good performance of noise removal, ADWA’s ability in signal’s feature preservation can be progressively improved by introducing new significant attributes; this makes ADWA can obtain an enough satisfactory performance in both noise removal and feature preservation. Therefore, ADWA provides an effective and promising way to cope with the denoising contradiction.