Abstract
Auto-detecting a submerged human body underwater is very challenging with the absolute necessity to a diver or a submersible. For the vision sensor, the water turbidity and limited light condition make it difficult to take clear images. For this reason, sonar sensors are mainly utilized in water. However, even though a sonar sensor can give a plausible underwater image within this limitation, the sonar image’s quality varies greatly depending on the background of the target. The readability of the sonar image is very different according to the target distance from the underwater floor or the incidence angle of the sonar sensor to the floor. The target background must be very considerable because it causes scattered and polarization noise in the sonar image. To successfully classify the sonar image with these noises, we adopted a Convolutional Neural Network (CNN) such as AlexNet and GoogleNet. In preparing the training data for this model, the data augmentation on scattering and polarization were implemented to improve the classification accuracy from the original sonar image. It could be practical to classify sonar images undersea even by training sonar images only from the simple testbed experiments. Experimental validation was performed using three different datasets of underwater sonar images from a submerged body of a dummy, resulting in a final average classification accuracy of 91.6% using GoogleNet.
Highlights
One dataset named CKI what stands for clean KIRO images, was captured in the very clean water testbed of Korea Institute of Robot and Convergence (KIRO)
All cases were expressed as the product of the classification accuracy for every 30 images of the body and background labels. Both AlexNet and GoogleNet models that trained with original CKI can see that there was no classification performance no matter what the model is because the average classification accuracy was lower than 50% ( 43.3% and 31.6% prospectively)
We confirmed that we can classify images captured from the ocean using only sonar images obtained from the testbed in training the deep learning model
Summary
The techniques of recognizing an object in turbid water and making a decision on the type of the object are essential factors for a diver or a submersible to support marine tasks [1,2]. Cho et al proposed a method to provide an automatic detection alarm indicating the presence of suspected underwater objects using a high-speed imaging sonar. The efficiency of the operation, such as discrimination and judgment, is low due to the extreme fatigue because the operator has to be continuously observing the optical or sonar image There are few such studies of how to automatically detect a submerged body using an underwater sonar image captured in poor water conditions [7,8]. For this reason, sonar sensors are mainly utilized in water [9,10]. The target background effect should be considered to overcome because it causes scattered noise or polarization noise in the sonar image
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
