In order to reduce the acquisition time for a 3D image, a different light source configuration can be used. For example, structured light is a technique based on triangulation, and consists of projecting a known light pattern on to the target. Then, with appropriate signal processing techniques, 3D images can be obtained studying the deformation of the original pattern [40, 41].
Commonly, systems based on structured light comprise a digital light projector and a camera, placed at a known distance apart. The light patterns can be manifold, and several examples can be found in the literature [42]. However, small patterns are preferable in
in Figure 2.21. The receiver was a CCD camera, placed at a known distance from the light source.
Figure 2.21. Schematic of stripe structured light. The light stripe was swept across the target surface through a light projector, and a CCD camera was placed at a known distance from the light source. From [43].
This system was used to obtain images of two objects in water, under different scattering conditions, and a few image examples are shown in Figure 2.22. The scattering was varied by adding milk to tap water, and no attenuation coefficients were provided.
Figure 2.22. Target scene in water at different concentrations of milk.
The first column shows the projector illuminating the wide field scene.
The second column shows the projector illuminating the targets with a single light stripe. The third column shows the depth of the target scene, obtained scanning the targets with a single light stripe and detecting the light scattered back from the target with a camera. The last column shows the reconstruction of the target reflectance. From [43].
The results show that the system was able to identify the targets also in highly scattering conditions (last row of Figure 2.22), and the reconstructed surface is shown in the point cloud graph in the third column. In addition, Narasimhan et al. developed an algorithm to recover the reflectance of the targets at different colour channels [43], and the results are shown in the last column of Figure 2.22. It is interesting to note that in highly scattering conditions, they were able to recover the depth and reflectance of most of the target scene, although the handle of the cup is missing because of the low signal detected due to the high level of attenuation. However, no details about the resolution were provided in the reference.
Structured light techniques can be improved by adding more receivers to the system, and in this case several configurations can be considered. For example, Bruno et al. in 2011 used two CCD cameras in conjunction with a projector, which illuminates the entire target with a programmable light pattern [44]. However, despite each pattern covering the entire
With this system, Bruno et al. were able to obtain images at stand-off distances up to approximately 4 attenuation lengths, and few results are shown in Figure 2.23. The figure shows in column a) the photographs of the target illuminated by one light pattern, in three different water conditions. While column b) in Figure 2.23 shows the point cloud graphs obtained by structured illumination technique.
Figure 2.23. (a) Photographs of one pattern projected on the target at different turbidity levels. (b) Point cloud graphs of the reconstructed target. The system comprised two CCD cameras and a light projector, programmed to project different light patterns. Adapted from [44].
Alternatively, a different source can be considered for structured light. For example, Massot-Campos et al. in 2014 used a CCD camera in conjunction with a green continuous wave laser and a diffractive optical element in front of the laser beam [45]. This system allowed for the use of multiple laser lines at the same time in order to form a structured illumination, as shown schematically in Figure 2.24. The laser used a wavelength of λ = 532 nm and 5 mW optical power, and it is shown in Figure 2.25a). The photograph
shows the laser in the white housing, while the camera is in the black housing. The system was tested in unfiltered tap water in the tank of the Ocean Systems Laboratory at Heriot-Watt University. The tank was 4 metres long, 3 metres wide, and 2 metres deep (Figure 2.25b).
Figure 2.24. Schematic of single frame structured light system based on triangulation. A light pattern is projected on the target, which deforms the original pattern. Studying the deformations, it is possible reconstruct a 3D map of the target. From [45].
Figure 2.25. (a) Photograph of the system, the black housing comprised a CCD camera, while the white housing comprised a laser with wavelength λ = 532 nm and 5 mW optical power. (b) The system was tested in the water tank of the Ocean Systems Laboratory at Heriot-Watt University.
Figure 2.26 shows an example of a non-processed 3D reconstruction of a 15 cm plastic weight plate in unfiltered tap water, (b) front view and (c) bottom view [45]. Preliminary tests showed that the laser based structured light system allowed sharp and clear details [46]. However, a quantitative study in clear and turbid water conditions was not yet performed and is part of future work.
Figure 2.26. (a) Laser input frame obtained with the system in Figure 2.25 illuminating a 15 cm plastic weight plate. (b) Front view of the 3D reconstruction of the weight plate. (c) Bottom view of the 3D reconstruction of the weight plate. From [45].