Share Email Print
cover

Proceedings Paper

Target detection for SAR imagery combining adaptive despeckling, region-based segmetation and shape matching
Author(s): Sang Hoon Lee
Format Member Price Non-Member Price
PDF $14.40 $18.00

Paper Abstract

The proposed target detection method for SAR imagery combines an adaptive despeckling technique, region-based image segmentation and shape matching. The radar wave coherence produces speckle in SAR imagery. This phenomenon gives to the images a granular appearance that complicates image analysis and interpretation in remote sensing tasks. The iterative maximum a posteriori(MAP) approach using a Bayesian model based on the lognormal distribution for image intensity and a Gibbs random field for image texture has been proposed for despeckling the SAR images that are corrupted by multiplicative speckle noise. However, the MAP estimation may use wrong information from adjacent different regions for the pixels located in the region close to or on the boundary. To overcome this problem, the new method is designed to utilize variable parameters depending on the pixel's location. The target objects should be isolated from the others within a given image. The despeckled image is then segmented into the objects. An object in the image can be represented by any measure based on the observed intensity values such as mean intensity and texture. Shape characteristics such as size and circularity can also represent an object. To segment the image, this study employs the RAG(regional adjacency graph)-based regiongrowing scheme using the local spatial and statistical information from the intensity values. Finally, the proposed method performs target detection by shape matching on the segmented objects. This study evaluated the new scheme using simulation data.

Paper Details

Date Published:
PDF
Proc. SPIE 7336, Signal Processing, Sensor Fusion, and Target Recognition XVIII, 73361N; doi: 10.1117/12.820173
Show Author Affiliations
Sang Hoon Lee, Kyungwon Univ. (Korea, Republic of)


Published in SPIE Proceedings Vol. 7336:
Signal Processing, Sensor Fusion, and Target Recognition XVIII
Ivan Kadar, Editor(s)

© SPIE. Terms of Use
Back to Top