Underwater visual position measurement system for high-accuracy beam installation
Introduction
In an underwater beam installation process, a computer-controlled position adjustment system is established to improve the efficiency of the construction process. In order to ensure the fitting accuracy of two beams in the fine adjustment and to achieve automatic control of the system, high accurate underwater position measurement is required. Among the various available solutions, underwater visual measurement provides higher accuracy and resolution compared with those based on radio and ultrasonic sources that cannot be adequately focused [1]. Monocular visual methods have been used for position measurement to reduce the computation time and operation cost but the depth information is unable to be obtained from one image [2]. Estimation of the depth will require extra information provided by a second image acquired through a known camera motion or by a priori knowledge related to geometric information extracted from the 2D image [3]. However, camera motion will cause greater variations in camera calibration parameters in underwater measurement because of radial distortion, which will affect the measurement accuracy. Moreover, radial supporting ways for moving cameras are a challenging issue [4]. Besides, underwater imaging always suffers from low contrast, non-uniform illumination or diminished colors, which makes feature extraction and object recognition from underwater images very difficult [5], [6], [7].
Compared with mono-vision, stereo vision can obtain the depth information directly from the images based on the measurement of parallax [8]. Ishibashi [9] described a stereo vision system to calculate three-dimensional position data of an object in the underwater working environment. Zheng et al. [10] presented an underwater stereo vision system to measure the distance between the target and the camera as well as the 3D information of the target. Nevertheless, it's very challenging to catch homologous points in the stereo pair of underwater images because of many interference factors such as photometric distortions and noise, occlusions and discontinuities, etc.[8]. Bruno et al. [11] used a Gray-code technique to solve the correspondence between the points in the underwater stereo pairs. In addition, stereo vision requires precise previous calibration for the extrinsic parameters of the stereo pair to achieve high accurate position measurement [12], [13].
This paper employs two pairs of bi-camera vision system to measure the position of a large-size cross beam to be installed referring to an installed one in an underwater operation. The bi-camera vision system rather than stereo vision can overcome the homologous point correspondence problem with stereo-based system and precise previous calibration for the extrinsic parameters of the stereo pair is no longer necessary. As it's difficult to extract feature points from the underwater image due to blur edges and unclear contours on the image, which results in low accuracy of position measurement underwater, super resolution as well as other image enhancement technique has been applied in this research to solve the problem.
Section snippets
Overall underwater position measurement system
As shown in Fig. 1, the overall underwater beam installation system employs two pairs of bi-camera vision system on both sides of the beam to measure the relative position of the beam to be installed referring to an installed one. Since the beams are of large size, it's difficult to obtain a full view of the underwater beams. Rails are fixed on the beams and appropriate markers, as the measuring objects, are coated on them, as shown in Fig. 2. In the bi-camera vision system, one camera faces
Image pre-processing
As is discussed above, super resolution method is used to restore the image with blur edges and unclear contours. First, an image degradation model [14] is adopted to analyze the problem of the degraded image, as the following shows:
Yk is the kth representative image of the high-resolution image X the low-resolution image frame of size N. The matrix Dk represents the decimation operation caused by underwater light attenuation. The matrix Ck represents the linear blur
Experiment set up
Experiments have been conducted in the Yangtze River, Wuhan. Four DH-SV401GC/GM digital cameras with the Kowa 16 mm lens are used. In the experiment, two aligned beams are first placed underwater. Three-way adjustment devices are then controlled to adjustment one beam to the set position. The set distance of horizontal spacing, vertical misalignment and height misalignment can be told from the elongation of the adjustment cylinder in the corresponding direction, which is measured by the
Conclusion
This paper focuses on an underwater visual position measurement system to ensure the fitting accuracy of two beams in the fine adjustment of an underwater beam installation process. Restricted by the limited observability range of the underwater cameras, appropriate markers (as the measuring objects) are coated on the rails fixed on the beam. Unlike the binocular stereo vision, here a bi-camera vision system consisting of a top-view camera and a side-view camera is put on each side of the beam
Acknowledgment
The work is supported by the Fundamental Research Funds for the Central Universities and National Key Technology Support Program under Grant No. 2015BAF07B05.
References (16)
- et al.
Design and analysis of radial support for a large-aperture rotating mirror
Optik
(2014) - et al.
Visual-adaption-mechanism based underwater object extraction
Opt. Laser Technol.
(2014) - et al.
Experimentation of structured light and stereo vision for underwater 3D reconstruction
ISPRS J. Photogramm. Remote Sens.
(2011) - et al.
A review of the status and trends in underwater videometric measurement. Invited paper
SPIE Conf.
(2007) - et al.
Monocular vision system for distance measurement based on feature points
Opt. Precis. Eng.
(2011) - et al.
Monocular visual self-localization for humanoid soccer robots. Electrical Communications and Computers (CONIELECOMP)
- et al.
Color-based underwater object recognition using water light attenuation
Intell. Serv. Robot.
(2012) - et al.
Exploring underwater target detection by imaging polarimetry and correlation techniques
Appl. Opt.
(2013)