Modified Non-Local Means De-Noising with Patch and Edge Patch based Dictionaries

Image denoising is a fundamental yet challenging problem that has been studied for decades. Dictionary is built only once with high resolution images belonging to different scenes. Since the dictionary is well organized in terms of indexing its entries, it is used to search similar patches very quickly for efficient NLM denoising.Our approach is very different in the sense, we build a single dictionary as a pre-processing step with a large number of arbitrary but high quality, textured images belonging to different object classes in contrast with building a dictionary for each noisy test image separately. We prefer to do this to avoid the time required to build the dictionary each time for separate test images.Principle Component Analysis (PCA) is a standard tool in modern data analysis because it is simple method for extracting relevant information from complex data matrix using eigen-values and eigenvectors. The multi-scale principal component generalizes the usual PCA of a multivariate signal seen as a matrix by performing simultaneously a PCA on the matrices of details of different levels. In multi scale Principal Component Analysis (MSPCA) de-correlate the variables by extracting a linear relationship and wavelet analysis. For dictionary building purpose smooth patches has been find using PCA. Then clustering is done for the similar patches. Then id is allocated for each cluster to get the dictionary.