Anne-Gaëlle Allais

3D Reconstruction of Natural Underwater Scenes Using the Stereovision System IRIS

The aim of this study is to propose a 3-dimension reconstruction method of small-scale scenes improved by a new image acquisition method for quantitative measurements. A stereovision system is used to acquire images in order to obtain several shots of an object, at regular intervals according to a predefined trajectory. A complete methodology of 3D reconstruction is exposed to perform a dense 3D model with texture mapping. A first result on natural images collected with the stereovision system during sea trials has been obtained.

Influence of polarization filtering on image registration precision in underwater conditions

Underwater images often suffer from poor visibility due to photon scattering. However, in some cases, optical polarization filtering techniques can decrease the contribution of the scattered light and improve the visual image quality. In this Letter, the influence of these techniques for underwater image registration is analyzed, particularly when backscattered light is the main perturbation induced by the submarine environment. This analysis is performed using the Cramer-Rao bound and relies on a standard image formation model, taking into account various kinds of noises.

Color image simulation for underwater optics

Underwater optical image simulation is a valuable tool for oceanic science, especially for the characterization of image processing techniques such as color restoration. In this context, simulating images with a correct color rendering is crucial. This paper presents an extension of existing image simulation models to RGB imaging. The influence of the spectral discretization of the model parameters on the color rendering of the simulated images is studied. It is especially shown that, if only RGB data of the scene chosen for simulations are available, a spectral reconstruction step prior to the simulations improves the image color rendering.

Active Stereovision Using Invariant Visual Servoing

The objective of this paper is to propose an innovative visual servoing method in order to improve the 3D reconstruction of objects for quantitative measurements. The method uses a Stereovision system that allows us to obtain several shots of an object, at regular intervals according to a predefined trajectory. In our case, the stereo rig is equipped with two different cameras, consequently the intrinsic parameters are not the same. The first one is fixed while the other one is mounted on a pan and tilt. Simulation and preliminary experiments in laboratory conditions shows the validity of our approach

Automatic classification of deep benthic habitats: Detection of microbial mats and siboglinid polychaete fields from optical images on the Håkon Mosby Mud Volcano

While many seafloor surveys provide a growing number of data, only few automatic techniques are developed to analyze images. This study is interested in the automatic detection and quantification of microbial mats and fields of siboglinid polychaete (tubeworm) colonizing the seafloor of the deep-sea Hakon Mosby Mud Volcano surveyed by ROV. Three algorithms are developed to segment the high resolution optical images of the seafloor which apply a watershed transformation coupled with a region growing technique, or a similarity measure operated with a region growing technique, or a texture analysis used with the Kullback-Leibler divergence. The results are compared through the score of a classification ratio estimated with a human-made classification.

Low overlap image registration based on both entropy and mutual information measures

In this paper, a new approach to image matching based on both entropy and mutual information measures that can correctly perform association under very low overlap conditions is presented. The method is feature-based, working on a list of interest points detected on the acquired images. Unlike the classical SSD/RanSaC method, our algorithm is robust to the existence of many similar regions in the images, enabling us to handle situations where the interest points correspond to local details of an habitat that is dispersed with a regular-like structure across the ocean bottom. Actually, our method requires only that four corresponding points be detected in the two images being aligned. This allows matching under minimal overlap of the images. The improvement in efficiency of our method, when compared to previous techniques, relies on explicitly accounting for ambiguities in the association between the templates of the two images, thus preventing that useful information be discarded at an early association step.