Miquel Massot-Campos170, Gabriel Oliver-Codina171
Structure from motion or stereoscopy are used to obtain 3D from a sequence of still
images. However if there is no texture or features in the images, no 3D can be obtained. Featureless environments are difficult to reconstruct in 3D only using cameras. Projected light patterns can be used to measure the shape or an object. However, scattering is the main problem in light based underwater sensors such as light
and cameras. Collimated light such as laser minimizes this problems by focusing the
light in fewer points. Ranged gated cameras can be also used with pulsed lasers to
reduce even more the scattering. In this paper a new structured light laser system is
evaluated to solve the scattering and featureless problems above mentioned and
to perform a 3D underwater reconstruction. This system is formed by a laser projector and a camera. By means of pattern identification and triangulation, 3D can be
reconstructed from a live video sequence.
Keywords – Structured Light, Laser Line Scanning, Underwater Sensors, 3D Reconstruction
Autonomous grasping of unknown objects by a robot is a challenging task. In
the last years, is receiving increasing attention in underwater environments.
These environments are highly unstructured and limit the availability and effective range of sensors. Grasping an object generally requires the knowledge of a
partial 3D structure or prior knowledge of the CAD model the robot is going to
manipulate [1]. In order to obtain these data different methods exist. They can
be generally classified according to the type of the sensing device: sonar, laser,
stereoscopy, structured light. Sonar based methods are the most extended underwater. These sensors can measure distances of hundreds of meters with a
resolution of a centimeter, which is not enough for manipulation. Laser rangefinders are common in wheeled robots, but they do not work correctly underwater due to energy absorption of water. Laser line scanning (LLS) is being used
to sweep an area while the robot is stationary and, with a camera, the 3D is
recovered [2]. This technique depends on a laser line projecting device and a
camera. With two cameras 3D can also be computed, however the density of the
reconstruction is directly related to the texture of the object. Finally, structured
light (SL) projects a pattern that creates artificial texture on the required object
in order to solve the stereoscopy problem [3].
Detailed 3D point clouds are needed to perform a correct manipulation or
grasping in underwater environments. The sensors that are accurate enough to
perform such a task are LLS, stereo cameras and SL. Stereo cameras need texture in order to compute 3D points, and in these environments the object to be
manipulated and the surrounding area where it is laying can be featureless because of mud or flora growing [4]. LLS is slow but accurate, whilst SL is fast but in
underwater environments suffer from scattering problems. Solutions to these
options have been explored by researchers in the last decade, and we would like
to introduce a new underwater sensor based on structured light systems and
laser light, which is under study.
The performance of underwater optical imaging systems such as cameras is
limited by absorption and scattering. These two terms depend on the turbidity of the water the light is propagating in. When a system has been optimized
to reduce backscatter it may become limited by absorption. In this situation,
the propagating signal is too weak to be detected by the corresponding sensor,
and the system is said to be power limited. If the power is increased the scattering increases. It can increase so much that the sensor cannot differentiate the
true signal from the noise. In this case the system is said to be contrast limited.
The performance in both cases can be enhanced by choosing the source wavelength to match the optimal underwater wavelength, which is in the blue-green
range of colors in the visual spectra. This enhances the range. If a collimated
light is also chosen, polarization filters can be also used to discard the light scat-
tered by suspended particles. Advanced systems reduce even more the scattering by controlling the temporal properties of light, gating both the light emitter
and the receiver [5].
Fig. 1 shows the proposed laser projector, projecting a 25 parallel line pattern
onto a textureless white plastic bottle. On the right image the camera frame
can be seen. With this system 3D pointcloud data can be provided at per frame
rate, without the need of waiting a laser sweep scan nor having the need of a
textured surface.
Until now, this system has been tested in an Underwater Simulator (UWSim,
http://www.irs.uji.es/uwsim/) and in a controlled environment.
[1] M. Prats, J. J. Fernandez, and P. J. Sanz, “An approach for semi-autonomous recovery of unknown objects in underwater environments,” in 2012 13th International
Conference on Optimization of Electrical and Electronic Equipment (OPTIM), 2012,
pp. 1452–1457.
[2] F. Dalgleish, F. Caimi, W. Britton, and C. Andren, “Improved LLS imaging performance in scattering-dominant waters,” Proc. SPIE, vol. 7317, pp. 1–12, 2009.
[3] S. G. Narasimhan and S. K. Nayar, “Structured Light Methods for Underwater Imaging: Light Stripe Scanning and Photometric Stereo,” Proceedings of OCEANS 2005
MTS/IEEE, pp. 1–8, 2005.
[4] T. D. Dao, “Underwater 3D reconstruction from stereo images,” University of Girona (Spain), University of Burgundy (France), Heriot Watt University (UK), 2008.
[5] F. Dalgleish, F. Caimi, and W. Britton, “An AUV-deployable pulsed laser line scan
(PLLS) imaging sensor,” IEEE Oceans, pp. 1–5, 2007.
Fig. 1: Laser based structured light projector.
Instrumentation Viewpoint / 15/ MARTECH 13
Download PDF