Open AccessFine logarithmic adaptive soft morphological algorithm for synthetic aperture radar image segmentation
Author(s) -
Koosha Mohaddeseh,
Hajsadeghi Khosrow,
Koosha Mohammad
Publication year - 2014
Publication title -
iet image processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.401
H-Index - 45
eISSN - 1751-9667
pISSN - 1751-9659
DOI - 10.1049/iet-ipr.2013.0046
Subject(s) - synthetic aperture radar , artificial intelligence , computer science , speckle noise , computer vision , image segmentation , filter (signal processing) , speckle pattern , inverse synthetic aperture radar , segmentation , median filter , radar imaging , noise (video) , image processing , scale space segmentation , pattern recognition (psychology) , image (mathematics) , radar , telecommunications
Synthetic aperture radar (SAR) appropriate image processing in conjunction with noise reduction is crucial in proper image segmentation. The authors present a new algorithm, logarithmic adaptive soft morphological (LASM) filter, utilising collectivity and flexibility of order‐statistic soft morphological filters. This method not only reduces the speckle noise of the single‐look SAR imagery considerably, but it significantly enhances the segmentation results. To verify the performance, a simulated SAR image is first created by applying an imagery method to an original noiseless image. The resulting image has characteristics identical to a real SAR image. The LASM method, as well as several well‐known and state‐of‐the‐art filtering algorithms, is then applied to the image to filter out the speckles. The LASM filtered results are then compared with other results by evaluating them against the original image. Some of these methods are also applied to a real SAR image and the results are also compared. Segmentation is then applied to the filtered images and the results are used as a deterministic evaluation metric. The results show marked improvement using the LASM processing the amplitude images compared with well‐known methods, whereas the superiority of LASM algorithm is even more pronounced in processing the intensity images.