APPLICATION OF IMPROVED SIGMA FILTERS TO POLARIMETRIC SAR IMAGES
| Autor: | Boutarfa, Souhila, Bouchemakh, Lynda, Smara, Youcef |
|---|---|
| Přispěvatelé: | Lacaze, Bernard |
| Jazyk: | francouzština |
| Rok vydání: | 2013 |
| Předmět: |
POLSAR images
improved Sigma filter filtre sigma amélioré [SHS.GEO] Humanities and Social Sciences/Geography Filtre sigma de Lee adaptive Lee filter speckle filtering images POLSAR filtre adaptatif de Lee polarimétrie filtre du chatoiement Lee Sigma filter polarimetry [SPI.SIGNAL] Engineering Sciences [physics]/Signal and Image processing |
| Popis: | The images acquired by a Polarimetric Synthetic Aperture Radar (POLSAR) are characterized by an inherentmultiplicative noise called speckle. It corrupts both the amplitude and the phase, which complicates data interpretation,degrades segmentation performance and reduces target detectability. Hence, we need to preprocess the images byadaptive filtering methods before their analysis.In this paper, we present the Lee Sigma filter and its two enhanced versions that are based on knowledge of the radarprobability density. The Sigma filter uses the calculation of the local mean from the pixels included in a two-Sigmainterval. The improved Sigma filter is based on the estimation of the Minimum Mean Square Error MMSE. Theimproved Sigma filter with strong scatterer detection is based on the point targets preservation and application of theMMSE on the pixels of image.The methods are applied on three polarimetric images (HH, HV, VV) acquired by the ESAR sensor corresponding tothe area located in Oberpfaffenhofen, in Munich, Germany, and on the bi - polarization images acquired by theENVISAT satellite corresponding to the region of Oran in Algeria, and on the fully polarimetric images acquired by theRADARSAT-2 satellite on the area of Algiers, Algeria. Les images acquises par les systèmes radar SAR polarimétrique (POLSAR) sont caractérisées par la présence d'un bruit nommé bruit de chatoiement ou speckle. Ce dernier, de nature multiplicative, corrompt à la fois l'amplitude des images et la phase, ce qui complique l'interprétation des données, dégrade la performance de segmentation et réduit la détectabilité des cibles. D'où la nécessité de prétraiter les images par des méthodes de filtrage adaptées, avant de procéder à leur analyse. Dans cet article, nous présentons le filtre Sigma de Lee et ses deux versions améliorées qui se basent sur la connaissance de la densité de probabilité du signal radar. Le filtre Sigma utilise le calcul de la moyenne locale à partir des pixels compris dans un intervalle de valeur égale à deux-Sigma. Le filtre Sigma amélioré repose sur l'estimation de l'erreur quadratique moyenne minimale EQMM (MMSE : Minimum Mean Square Error). Le filtre Sigma amélioré avec détection de forts diffuseurs se base sur la préservation des points fortement réfléchis et l'application de l'EQMM sur les pixels de l'image. Les méthodes sont appliquées sur trois images polarimétriques (HH, HV, VV) acquises par le capteur ESAR correspondant à la zone d'Oberpfaffenhofen, de Munich en Allemagne, sur des images d'Oran en Algérie en bi-polarisations acquises par le satellite ENVISAT, ainsi que sur des images totalement polarimétriques d'Alger, Algérie, acquises par le satellite RADARSAT-2. Mots clés : Images POLSAR, filtrage de chatoiement, polarimétrie, filtre de Lee adaptatif, filtre Sigma de Lee, filtre Sigma amélioré. Abstract The images acquired by a Polarimetric Synthetic Aperture Radar (POLSAR) are characterized by an inherent multiplicative noise called speckle. It corrupts both the amplitude and the phase, which complicates data interpretation, degrades segmentation performance and reduces target detectability. Hence, we need to preprocess the images by adaptive filtering methods before their analysis. In this paper, we present the Lee Sigma filter and its two enhanced versions that are based on knowledge of the radar probability density. The Sigma filter uses the calculation of the local mean from the pixels included in a two-Sigma interval. The improved Sigma filter is based on the estimation of the Minimum Mean Square Error MMSE. The improved Sigma filter with strong scatterer detection is based on the point targets preservation and application of the MMSE on the pixels of image. The methods are applied on three polarimetric images (HH, HV, VV) acquired by the ESAR sensor corresponding to the area located in Oberpfaffenhofen, in Munich, Germany, and on the bi-polarization images acquired by the ENVISAT satellite corresponding to the region of Oran in Algeria, and on the fully polarimetric images acquired by the RADARSAT-2 satellite on the area of Algiers, Algeria. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|