Development of a machine learning technique for automatic analysis of seafloor image data: Case example, Pogonophora coverage at mud volcanoes

Abstract

a b s t r a c tDigital image processing provides powerful tools for fast and precise analysis of large image data sets in marine and geoscientific applications. Because of the increasing volume of georeferenced image and video data acquired by underwater platforms such as remotely operated vehicles, means of automatic analysis of the acquired image data are required. A new and fast-developing application is the combination of video imagery and mosaicking techniques for seafloor habitat mapping. In this article we introduce an approach to fully automatic detection and quantification of Pogonophora coverage in seafloor video mosaics from mud volcanoes. The automatic recognition is based on textural image features extracted from the raw image data and classification using machine learning techniques. Classification rates of up to 98.86% were achieved on the training data. The approach was extensively validated on a data set of more than 4000 seafloor video mosaics from the Håkon Mosby Mud Volcano.

Evaluation of seabed mapping methods for fine-scale classification of extremely shallow benthic habitats – Application to the Venice Lagoon, Italy

“…Similarly to others (i.e. Lüdtke et al 2012;Stephens and Diesing, 2014), we noticed that the KNN performs well when the image is preceded by small scale multiresolution segmentation. However, our investigation shows that increasing the K parameter decreased diversity of class distribution, resulting in a less accurate classification.…”

Spatial scale and geographic context in benthic habitat mapping: review and future directions

“…Of interest are developments in automatic species detection and analysis on video data (e.g. Purser et al 2009, Lüdtke et al 2012, Seiler et al 2012, Tanner et al 2015, in methods for generating photo-mosaics of the seafloor for accurate georeferencing (e.g. Prados et al 2012, Kwasnitschka et al 2013, Marsh et al 2013, in spatial statistics (e.g.…”

Assessment of Machine Learning Algorithms for Automatic Benthic Cover Monitoring and Mapping Using Towed Underwater Video Camera and High-Resolution Satellite Images

“…However, the analysis of recorded towed videos in marine applications is usually performed manually [3], and automatic feature extraction is not often applied [4,5]. Therefore, the automatic classification of benthic habitats from towed underwater photos is a comparatively novel and innovative method [6].…”

“…Examples of attempts used for benthic cover detection with underwater video systems can be found in the literature [7]. However, most researchers process photos captured from towed cameras mounted on remotely operated vehicles (ROVs) or autonomous underwater vehicles (AUVs) [3]. Adam et al [2] investigated a random forest, neural network, and classification trees machine learning algorithms to classify two seabed categories, sand and maerl, using (RGB or LAB) pixels manually extracted from images captured by an ROV.…”