Underwater 3D Object Reconstruction for Fish Length Estimation Using Convolutional Neural Networks

Ubina, Naomi A.; Cai, Sin-Yi; Cheng, Stone; Chang, Chin-Chun; Hsieh, Yi-Zeng

doi:10.1109/ispacs51563.2021.9651057

Cited by 7 publications

(1 citation statement)

References 3 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Direct training with unprocessed data significantly reduces the training process’s complexity. To validate the procedure, underwater fish reconstruction experiments were performed employing this approach; the outcomes indicate that the error rate remains below 6% [ 23 ]. Nocerino et al introduced a dense image-matching technique along with a thorough evaluation and analysis of the chosen method.…”

Section: Introductionmentioning

confidence: 99%

A Study on Refraction Error Compensation Method for Underwater Spinning Laser Scanning Three-Dimensional Imaging

Zhang,

Wang,

Zhang

et al. 2024

Sensors

View full text Add to dashboard Cite

Laser scanning 3D imaging technology, because it can obtain accurate three-dimensional surface data, has been widely used in the search for wrecks and rescue operations, underwater resource development, and other fields. At present, the conventional underwater spinning laser scanning imaging system maintains a relatively fixed light window. However, in low-light situations underwater, the rotation of the scanning device causes some degree of water fluctuation, which warps the light strip data that the system sensor receives about the object’s surface. To solve this problem, this research studies an underwater 3D scanning and imaging system that makes use of a fixed light window and a spinning laser (FWLS). A refraction error compensation algorithm is investigated that is based on the fundamentals of linear laser scanning imaging, and a dynamic refraction mathematical model is established based on the motion of the imaging device. The results of the experiment on error analysis in an optimal underwater environment indicate that the error in reconstructing the radius is decreased by 60% (from 2.5 mm to around 1 mm) when compensating for the measurement data of a standard sphere with a radius of 20 mm. Moreover, the compensated point cloud data exhibit a higher degree of correspondence with the model of the standard spherical point cloud. Furthermore, we examine the impact of physical noise, measurement distance, and partial occlusion of the object on the imaging system inside an authentic underwater setting. This study is a good starting point for looking at the refractive error of an underwater laser scanning imaging system. It also provides to us some ideas for future research on the refractive error of other scanning imaging methods.

show abstract

Section: Introductionmentioning

confidence: 99%