Abstract
Images acquired by synthetic aperture radars are degraded by speckle that prevents efficient extraction of useful information from radar remote sensing data. Filtering or despeckling is a tool often used to improve image quality. However, depending upon image and noise properties, the quality of improvement can vary. Besides, a quality can be characterized by different criteria or metrics, where visual quality metrics can be of value. For the case study of discrete cosine transform (DCT)based filtering, we show that improvement of radar image quality due to denoising can be predicted in a simple and fast way, especially if one deals with particular type of radar data such as images acquired by Sentinel-1. Our approach is based on application of a trained neural network that, in general, might have a different number of inputs (features). We propose a set of features describing image and noise statistics from different viewpoints. From this set, that contains 28 features, we analyze different subsets and show that a subset of the 13 most important and informative features leads to a very accurate prediction. Test image generation and network training peculiarities are discussed. The trained neural network is then tested using different verification strategies. The results of the network application to test and real-life radar images are presented, demonstrating good performance for a wide set of quality metrics.
Highlights
Radar imaging has become an efficient tool for solving different monitoring tasks in forestry, hydrology, agriculture and many other applications [1,2,3,4]
Performance of many quality metrics has not been analyzed for the cases of pure multiplicative noise and residual distortions after filtering images originally distorted by speckle
peak signal-to-noise ratio (PSNR) and its modification, PSNR-HVS-M [47], that takes into account peculiarities of human vision system (HVS). Both metrics are expressed in dB; larger values correspond to better quality, metric values are positive; Visual quality metrics resulted from SSIM, such as FSIM [29], MSSSIM [50], IW-SSIM [51], ADD-SSIM [52], ADD-GSIM [52], SSIM4 [49]. All these metrics vary in the limits from 0 to 1; larger values relate to a higher visual quality; Visual quality metric WSNR [53] that is expressed in dB; it has positive values and larger ones relate to better visual quality; The recently proposed metric HaarPSI [54] varies from 0 to 1, having larger values for better quality images; The visual quality metric GMSD [48] is positive and smaller is better; The metric MAD [28] varies in wide limits, is positive and smaller is better; The metric GSM [55] varies in narrow limits, is smaller than unity and the larger the better; The metric DSS [56] varies in the limits from 0 to 1 and the larger the better
Summary
Radar imaging has become an efficient tool for solving different monitoring tasks in forestry, hydrology, agriculture and many other applications [1,2,3,4]. Performance of many quality metrics has not been analyzed for the cases of pure multiplicative noise and residual distortions after filtering images originally distorted by speckle. Keeping this in mind, we carry out further analysis for a limited number of conventional and visual quality metrics. All these metrics vary in the limits from 0 to 1; larger values relate to a higher visual quality; Visual quality metric WSNR [53] that is expressed in dB; it has positive values and larger ones relate to better visual quality; The recently proposed metric HaarPSI [54] varies from 0 to 1, having larger values for better quality images; The visual quality metric GMSD [48] is positive and smaller is better; The metric MAD [28] varies in wide limits, is positive and smaller is better; The metric GSM [55] varies in narrow limits, is smaller than unity and the larger the better; The metric DSS [56] varies in the limits from 0 to 1 and the larger the better
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have