摘要
针对水下单目非接触测量鱼体大小问题,通常需要测量水下鱼体距离获取其表面的三维信息,提出了一种基于线激光三角测距法的鱼体测距方案,并对水下激光测距技术展开研究。根据光路原理测距需要测量多个参数,过程复杂且误差大,计算速度慢。本研究测距方案首先通过张正友标定法实现了图像的畸变矫正还原了图像的真实性。使用OpenCv开源库识别鱼体表面的激光轮廓,自定算法提升了轮廓识别的准确率,对激光轮廓进行重心分析法并得出激光亮度重心坐标。通过分析激光图像点坐标与空间坐标的位移关系,拟合一元五阶多项式回归模型预测距离,简化了参数。结果显示:激光重心识别并不连续,但在鱼体身上都能准确识别。测距模型平均误差5.6 mm,测试的样本中80%预测结果的误差都在7.0 mm内,最大误差也在标定的2 cm内,能够满足鱼体测距的误差大小。研究表明,本方法可为进一步研究鱼体尺寸大小的表面三维信息获取提供参考。
For the problem of underwater single-eye non-contact measurement of fish size,it is usually necessary to measure the distance of underwater fish to obtain the three-dimensional information of its surface,this paper proposes a fish distance measurement scheme based on the laser triangulation method and conducts research on underwater laser ranging technology.According to the principle of optical path,multiple parameters need to be measured for ranging,which makes the process complex,error-prone,and slow in calculation.This paper first corrects the distortion of the image and restores its authenticity through Zhang Zhengyou’s calibration method.The OpenCv open-source library is used to recognize the laser contour on the surface of the fish body,and a self-defined algorithm improves the accuracy of contour recognition.Centroid analysis is applied to the laser contour to obtain the centroid coordinates of laser brightness.By analyzing the displacement relationship between the coordinates of the laser image points and the spatial coordinates,a polynomial regression model with a one-variable fifth-order is fitted to predict the distance,simplifying the parameters.The experimental results indicate that the laser center of gravity recognition is not continuous,but can accurately identify on the fish body.The average error of the ranging model is 5.6 mm.In 80%of the tested samples,the error of the predicted results is within 7.0 mm,and the maximum error is within 2 cm of calibration,which meets the required accuracy for fish body distance measurement.The research suggests that this method can provide reference for further acquisition of surface three-dimensional information of fish body size.
作者
俞圣池
李佳康
熊鑫泉
贺刘刚
何瑞麟
戴阳
YU Shengchi;LI Jiakang;XIONG Xingquan;HE Liugang;HE Ruilin;DAI Yang(College of Information,Shanghai Ocean University,Shanghai,201306,China;East China Sea Fisheries Research Institute,Chinese Academy of Fishery Sciences,Key Laboratory of Fisheries Remote Sensing,Ministry of Agriculture and Rural Affairs,P.R.China,Shanghai,200090,China;Laoshan Laboratory,Qingdao,266237,Shandong,China)
出处
《渔业现代化》
CSCD
北大核心
2024年第1期80-89,共10页
Fishery Modernization
基金
上海市2020年度科技创新行动计划(20dz1206400)
崂山实验室科技创新项目(LSKJ202201801)。
关键词
激光测距
测距法
三角测距法
光斑
鱼体测距
laser ranging
ranging method
triangulation
spot
fish body ranging
