Kernel Entropy Component Analysis Pre-Images for Pattern Denoising

Publication details

Part of: Image Analysis. 16th Scandinavian Conference, SCIA 2009 Oslo Norway, June 2009. Proceedings (Springer, 2009)
Pages: 626–635
Year: 2009
Links:
- SAMMENDRAG: http://www.springer.com/computer/image+processing/book/978-3-642-02229-6?changeHeader
- OMTALE: http://www.springer.com/computer/image+processing/book/978-3-642-02229-6?changeHeader

The recently proposed kernel entropy componenet analysis (kernel ECA) technique may produce strikingly different sprectral data sets than kernel PCA for a wide range of kernel sizes. In this paper, we investigate the use of kernel ECA as a componenet in a denoising technique previously developed for kernel PCA. The method is based on mapping noisy data to a kernal feature space, for then to denoise by projecting onto a kernel ECA subspace. The denoised data in the input space is obtained by computing pre-images of kernel ECA denoised patterns. The denoising results are in several cases improved.