A semi-automatic approach for estimating bedrock and surface layers from multichannel coherent radar depth sounder imagery

Abstract

The dynamic responses of the polar ice sheets in Greenland and Antarctica can have substantial impacts on sea level rise. Understanding the mass balance requires accurate assessments of the bedrock and surface layers, but identifying each layer is performed subjectively by time-consuming, dense hand selection. We have developed an approach for semi-automatically estimating bedrock and surface layers from radar depth sounder imagery acquired from Antarctica. Our solution utilizes an active contours method ("level sets") to propagate an initial estimation of a layer's position based upon curvature and image intensity gradients. This allows the initial curve to gravitate with topological changes while providing smooth boundaries for discriminating between bedrock and surface layers. We evaluated the proposed semi-automatic method on 20 images with respect to hand labeled ground-truth. Compared to an automatic technique, our approach reduced labeling error by factors of 5 and 3.5 for tracing bedrock and surface layers, respectively.

End-to-End Classification Network for Ice Sheet Subsurface Targets in Radar Imagery

“…Newer work in this field utilizes image processing, computer vision, and deep learning techniques to automatically or semi-automatically determine ice surface and bottom boundaries from echograms [1][2][3][4][5][6][7]. Gifford et al [1] employs both the edge-based and active contour methodologies to automate the task of locating polar ice and bedrock layers from airborne radar data acquired over Greenland and Antarctica.…”

“…A third technique, the level-set model, is better at identifying curved boundaries by embedding the evolving curve into a higher dimensional surface. Mitchell et al [2] uses a level-set technique for estimating bedrock and surface layers, but find it problematic because of the need to reinitialize the curve manually for each radar image. Therefore, Rahnemoonfar et al [3] introduces the distance regularization term, which essentially maintains the regularity of the level-set and leads to a stable numerical procedure without the need for reinitialization.…”

AI Radar Sensor: Creating Radar Depth Sounder Images Based on Generative Adversarial Network

“…Several semi-automated and automated methods exist for layer finding and estimating ice thickness in radar images [1][2][3][4][5][6]. Crandall et al [1] used probabilistic graphical models for detecting the ice layer boundary in echogram images from Greenland and Antarctica.…”

“…The extension of this work was presented in [2] where they used Markov-chain Monte Carlo to sample from the joint distribution over all possible layers conditioned on an image. Mitchell et al [3] used a level set technique for estimating bedrock and surface layers. However, for every single image, the user needs to re-initialize the curve manually and as a result, the method is quite slow and was applied only to a small dataset.…”

Estimating bedrock and surface layer boundaries and confidence intervals in ice sheet radar imagery using MCMC

“…A few recent papers have studied how to use image processing and computer vision techniques to determine layer boundaries automatically or semi-automatically from echograms [2,3,4,5,6,7,8,9,10], but this is a hard problem because of the high degree of noise, the often faint layer boundaries, and confusing linear structures caused by signal reflections and clutter. In fact, even human annotators produce diverging estimates of the boundaries in many cases.…”