Although there are numerous methods in the literature for removing noise from images, finding new methods remains a challenge in the field, and many of the existing methods, despite their complexity, do not achieve a sufficient level of applicability, most often due to the relatively long calculation time. Furthermore, most existing methods work well when the processed image is adapted to the algorithm, but fail or produce significant artifacts otherwise. Because the context of removing noise from images is similar to that of improving images, some concepts must be considered. Recognize that the proposed method will be used again. Soft truncation of the wavelet coefficients with threshold values adapted to the image's local statistics and correction based on the hierarchical correlation map are proposed as an adaptive spatial filter in the wavelet domain. The filter takes advantage of both the interband and bandwidth dependence of the wavelet coefficients in a novel way, while minimizing computational resources. The proposed method employs soft truncation of coefficients, in which the threshold value for each coefficient is calculated separately based on local statistics in the wavelet domain.
Author(s) Details
Faculty of Electrical Engineering and Computer Science, “Stefan cel Mare” University of Suceava, 720229 Suceava, Romania and National R & D Institute for Cryogenic and Isotopic Technologies, 240050 Rm. Valcea, Romania and Faculty of Building Services, Technical University of Cluj Napoca, 400114 Cluj Napoca, Romania.
Catalin Dumitrescu
Faculty of Transports, “Polytechnic” University of Bucharest, 060042 Bucharest, Romania.
Constantin Filote
Faculty of Electrical Engineering and Computer Science, “Stefan cel Mare” University of Suceava, 720229 Suceava, Romania.
Ioana Manta
National R & D Institute for Cryogenic and Isotopic Technologies, 240050 Rm. Valcea, Romania and Faculty of Power Engineering, “Polytechnic” University of Bucharest, 060042 Bucharest, Romania.
View Book :- https://stm.bookpi.org/RDER-V12/article/view/1484
No comments:
Post a Comment