Super-resolution of mammograms based on analysis of wavelet family and iterative scales
Juan Israel Yáñez-Vargas, Sheila Esmeralda Gonzalez-Reyna, Andrea González-Ramírez, Itzel Guerrero-Gasca, Felipe Astudillo-Montenegro · 2017
In this paper we introduce a super-resolution (SR) technique for enhanced the resolution in digital mammograms and increase details for future applications (classification), the method is based on a multi-scale iterative resolution (MSIR) and the wavelet transform (WT). The low-resolution digital mammography in first term is reconstructed in each nested refined SR frame via the iterative reconstruction of the upscaled images, followed by the discrete wavelet transform (twelve wavelet families are tested during all the process) promoting consistency preservation in each resolution frame, in each iteration we applied the denoising operator (wavelet thresholding) via soft thresholding. We feature the differences between the proposed MSIR and the most prominent Papoullis-Gerchberg SR technique adapted for the feature enhanced digital mammography and demostrate the advantages of the MSIR approach with several test with original mammograms.