Dynamic Range Compression Self-Adaption Method for SAR Image Based on Deep Learning

The visualization of synthetic aperture radar (SAR) images involves the mapping high dynamic range (HDR) amplitude values to gray levels for lower (LDR) display devices. This compression process determines visibility details in displayed result. It therefore plays a critical role remote sensing applications. There are some problems with existing methods, such as poor adaptability, detail loss, imbalance between contrast improvement and noise suppression. To effectively obtain suitable human observation subsequent interpretation, we introduce novel self-adaptive SAR image method based on deep learning. Its designed objective is present maximal amount information content eliminate contradiction noise. Considering that, propose decomposition-fusion framework. input rescaled certain size then put into bilateral feature enhancement module remap low frequency features realize suppression enhancement. Based features, fusion employed integration optimization achieve more precise reconstruction Visual quantitative experiments synthesized real-world show that proposed notably realizes which exceeds several statistical methods. has good adaptability can improve images’ interpretation.