Image Denoising Using Kernel Principal Component Analysis with Various Pre-imaging Techniques

Noise removal is a crucial step in image processing. Kernel Principal Component Analysis is an extensively used method for image denoising. The pre-image problem is a vital step in any denoising algorithm. In this paper, various pre-imaging techniques are discussed. It is also examined how the denoising performance is enhanced due to change in projection operation by applying the modified projection operation in one of the pre-imaging methods. With the help of toy examples and two image datasets, all the methods are discussed briefly.