Operative considerations
The principle of underwater photogrammetry does not differ from that of terrestrial or aerial photogrammetry, but it is necessary to take into account certain elements that may cause disturbances; in particular, the refraction of the diopter water-glass and the presence of the camera housing [39].
The specific constraints of the underwater medium (turbidity of water, presence of suspended particles) force the operators to work on a large scale, close to the objects (between 0.5 and 2 to 3 meters, depending on the water quality). This apparently constraining aspect imposes having to produce a great quantity of stereo pairs, but on the other hand it offers a very high degree of accuracy.
The important advantage of using photogrammetry in underwater surveys in comparison with the use of other techniques consists in its simplicity of implementation and the diversity of potential results (3D measurements, 3D reconstruction, orthophotography, and vector restitution).
The implementation only requires the use of a scale bar to compute the scale of the model. Moreover, if two or three synchronized cameras are used, additional equipment is not needed at the scene as the scale is computed using the calibration of the camera set. This approach also provides a relevant appreciation of the uncertainty of measurements; where, in addition, the photographs have to be taken with an important overlap. The key factor of this method is redundancy: each point of measured space must be seen in at least three photographs.
The operative advantage is related to the simplicity of the survey. Moreover, a submarine pilot can drive a remotely operated underwater vehicle (ROV) without having to undergo a long preliminary training period. This method requires little time and does not require specific personnel, thus greatly reducing the expenses in a context where time and costs of intervention are extremely high.
Camera calibration
Camera calibration in multimedia photogrammetry is a problem identified since almost 50 years [9; 23] . The problem has no obvious solution, since the light beam refraction through the different media (water, glass, air) introduces a refraction error which is impossible to express as a function of the image plane coordinates alone [37] . Therefore the deviation due to refraction is close to that produced by radial distortion even if radial distortion and refraction are two physical phenomena of different nature. For this reason, the approach described by Kwon [45] has been adopted, consisting in the use of standard photogrammetric calibration software to perform the calibration of the set housing + digital camera. This approach can indeed correct in a large part the refraction perturbation; however, it is strongly dependent on the optical characteristics of the water/glass interface of the housing. For a more rigorous approach, we can read the interesting developments made by Gili Telem, and Sagi Filin on underwater camera calibration
Automatic photogrammetry survey
The photogrammetric process is a very efficient procedure consisting mainly of three phases. The first phase is data acquisition by photographs which requires light processing. This process is non- intrusive (remote sensing), and necessitates only slightly time- consuming (only the time necessary to take pictures), and potentially a quite thorough practice. The second phase involves further data processing and is carried out in a laboratory. This phase, which is mainly automated, includes homologous point determination and pose estimation. The last phase, data interpretation and linking with domain knowledge (underwater archaeology for example) is always manual, performed by experts and very time-consuming.
SIFT algorithm is often used to determine the homologous points
[39, 40] and recently the FAST [41] algorithm is applied. Then the pose estimation process from relative orientation of stereo pair is obtained by the Stewenius algorithm [42; 46; 60] . The SBA open source software by Manolis Lourakis [51] and Noah Snavely [58] is applied for the global bundle adjustment. Finally several approaches are proposed for surface densification PMVS by Ponce and Furukawa [54; 55] . For a good overview of these techniques it is possible to refer to these paper [64] [40] . In our application, we chose three tools to solve this problem and we have developed some of them. We also develop bridge between them in order to take benefit of the best of each of them.
Underwater 3D survey merging optic and acoustic sensors
Optic and acoustic data fusion is an extremely promising technique for mapping underwater objects that has been receiving increasing attention over the past few years [53]. Generally, bathymetry obtained using underwater sonar is performed at a certain distance from the measured object (generally the seabed) and the obtained cloud point density is rather low in comparison with the one obtained by optical means.
Since photogrammetry requires working on a large scale, it therefore makes it possible to obtain dense 3D models. The merging of photogrammetric and acoustic models is similar to the fusion of data gathered by a terrestrial laser and photogrammetry. The fusion of optical and acoustic data involves the fusion of 3D models of very different densities – a task which requires specific precautions [44; 56] .
Only a few laboratories worldwide have produced groundbreaking work on optical/acoustic data fusion in an underwater environment. See for example [38] and [41] where the authors describe the use of techniques that allow the overlaying of photo mosaics on bathymetric 3D digital terrain maps [52] . In this case we have important qualitative information coming from photos, but the geometric definition of the digital terrain map comes from sonar measurements.
Optical and acoustic surveys can also be merged using structured light and high frequency sonar as by Chris Roman and his team [50]. This approach is very robust and accurate in low visibility conditions but does not carry over qualitative information.
No comments:
Post a Comment