农业机械学报2025，Vol.56Issue(10)：36-44,9.DOI:10.6041/j.issn.1000-1298.2025.10.003
基于双目立体视觉的冷冻黄鳍金枪鱼体尺与体质量测量方法
Binocular Stereo Vision-based Measurement Method for Body Size and Body Weight Measurement of Frozen Yellowfin Tuna
摘要
Abstract
Aiming to address the issues of significant measurement errors,low operational efficiency,and safety hazards inherent in the traditional manual weighing and measurement process for frozen yellowfin tuna(Thunnus albacares),a fish body dimension and body weight estimation method was proposed based on binocular stereo vision and the segment anything model(SAM)image segmentation algorithm.Using frozen yellowfin tuna in real processing environments as the research subject,the method comprised four modules:fish body target detection,fish body image segmentation,fish body dimension extraction,and fish body weight estimation modeling.The method involved capturing images of yellowfin tuna by using a depth camera.To tackle the challenge of complex environments in aquatic product processing settings,the YOLO v8n algorithm was employed to detect frozen yellowfin tuna,coupled with the BotSort object tracking algorithm and a dynamic mask algorithm to eliminate interference from extraneous objects.The SAM image segmentation algorithm was then used to precisely extract the image and depth information of the frozen yellowfin tuna.Combined with the Open3D framework,a 3D point cloud map and dimensional information of the fish were generated.Based on the extracted dimensions and a body weight estimation model,the method achieved accurate body weight prediction for frozen yellowfin tuna.The measured body dimensions and body weight data exhibited small errors:the mean absolute percentage error(MAPE)for body length was 4.35％,for body height was 3.37％,the mean absolute error(MAE)for estimated body weight was 0.604 kg,and the mean absolute percentage error was 4.63％.This method can efficiently and accurately achieve 3D measurement and body weight estimation of yellowfin tuna,providing a reference for the automated management of frozen yellowfin tuna processing.关键词
双目视觉/体尺测量/体质量测量/图像分割/三维点云/黄鳍金枪鱼Key words
binocular vision/body size measurement/body weight measurement/image segmentation/three-dimensional point cloud/yellowfin tuna分类
信息技术与安全科学引用本文复制引用
刘雨青,张昕程,万家城,王宸烨,隋辰曦,隋恒寿,万荣..基于双目立体视觉的冷冻黄鳍金枪鱼体尺与体质量测量方法[J].农业机械学报,2025,56(10):36-44,9.基金项目
国家重点研发计划项目(2023YFD2401301、2023YFD2401305) （2023YFD2401301、2023YFD2401305）
