Target Detection Based on High-Level Image Information for High-Resolution SAR Images

Abstract

With the rapid development of sensor technology, the higher resolution SAR images we can acquire. Therefore, we pay attention to not only the low-level image information but also the high-level image information when we detect target. Due to the multiplicative speckle noise largely interferes with its use, active contour model (ACM) is not appropriate for the target detection in SAR images, but we can make the most of the high-level image information (contour) offered by the method for target detection. Therefore, we introduce a target detection method based on ACMs in the paper. Two groups of comparison experiments show that the proposed method not only overcomes the difficulties that traditional ACMs are applied in target detection for SAR images, but also outperforms classical Markov random field (MRF) model in terms of accuracy. Besides, the proposed method is appropriate for the design of the SAR automatic target recognition (ATR) because of the use of ACM.