Search for: All records

Underwater image enhancement is often perceived as a disadvantageous process to object detection. We propose a novel analysis of the interactions between enhancement and detection, elaborating on the potential of enhancement to improve detection. In particular, we evaluate object detection performance for each individual image rather than across the entire set to allow a direct performance comparison of each image before and after enhancement. This approach enables the generation of unique queries to identify the outperforming and underperforming enhanced images compared to the original images. To accomplish this, we first produce enhanced image sets of the original images using recent image enhancement models. Each enhanced set is then divided into two groups: (1) images that outperform or match the performance of the original images and (2) images that underperform. Subsequently, we create mixed original-enhanced sets by replacing underperforming enhanced images with their corresponding original images. Next, we conduct a detailed analysis by evaluating all generated groups for quality and detection performance attributes. Finally, we perform an overlap analysis between the generated enhanced sets to identify cases where the enhanced images of different enhancement algorithms unanimously outperform, equally perform, or underperform the original images. Our analysis reveals that, when evaluated individually, most enhanced images achieve equal or superior performance compared to their original counterparts. The proposed method uncovers variations in detection performance that are not apparent in a whole set as opposed to a per-image evaluation because the latter reveals that only a small percentage of enhanced images cause an overall negative impact on detection. We also find that over-enhancement may lead to deteriorated object detection performance. Lastly, we note that enhanced images reveal hidden objects that were not annotated due to the low visibility of the original images. 

Predicting the channel quality for an underwater acoustic communication link is not a straightforward task. Previous approaches have focused on either physical observations of weather or engineered signal features, some of which require substantial processing to obtain. This work applies a convolutional neural network to the channel impulse responses, allowing the network to learn the features that are useful in predicting the channel quality. Results obtained are comparable or better than conventional supervised learning models, depending on the dataset. The universality of the learned features is also demonstrated by strong prediction performance when transferring from a more complex underwater acoustic channel to a simpler one. 

The propagation of acoustic waves under water is a highly complex and stochastic process. Such channel dynamics renders large performance variation in underwater acoustic (UWA) communications. Prediction of the UWA communication performance is critical for selection and adaptation of the communication strategies. This work explores the use of supervised learning for performance prediction in UWA communications. This work first quantifies the transmitter design, the UWA channel characteristics and the receiver design by numerical and categorical parameters. For a chosen performance metric (e.g., the bit error rate or the packet error rate), the performance prediction is cast individually into a numerical prediction problem and a classification problem. Using the data sets from two field experiments, the performance of typical supervised learning methods are examined. The data processing results reveal that some supervised learning methods can achieve fairly good numerical prediction or classification performance, and the discriminative models typically outperform the generative models.