基于三维重建的猪大排定量切片方法  

Quantitative slicing method based on three-dimensional reconstruction of porcine large chops

作　　者：江政 俞驰威 黄天然 缪佳晖 陈坤杰[1] 黄明[3] 黄继超 JIANG Zheng;YU Chiwei;HUANG Tianran;MIAO Jiahui;CHEN Kunjie;HUANG Ming;HUANG Jichao(College of Engineering,Nanjing Agricultural University,Nanjing 210031,China;Jiangsu Livestock and Poultry Products Processing Engineering and Technology Research Center,Nanjing Professor Huang Food Science and Technology Company,Nanjing 211225,China;College of Food Science and Technology,Nanjing Agricultural University,Nanjing 210095,China)

机构地区：[1]南京农业大学工学院,南京210031 [2]江苏省畜禽产品加工工程技术研究中心,南京黄教授食品科技有限公司,南京211225 [3]南京农业大学食品科学技术学院,南京210095

出　　处：《农业工程学报》2025年第2期291-299,共9页Transactions of the Chinese Society of Agricultural Engineering

基　　金：“十四五”国家重点研发计划项目(2022YFD2100502);江苏省科技项目(BE2023307)。

摘　　要：针对猪大排定量切片准确度低的问题,该研究提出了一种基于三维重建的猪大排定量切片方法。首先,通过搭建点云数据采集试验平台,利用线激光扫描仪采集不同视角下的猪大排点云数据,并经过点云预处理提取目标点云;其次,结合长方体标靶块和Trimmed icp算法完成多视角的点云配准工作;然后,利用Graham扫描法以及不规则体切片分割累加算法估算猪大排的体积;最后根据设定的切片质量要求进行切片厚度计算,完成定量切片试验。采用15条猪大排样本对该研究所提出的方法进行验证,结果表明,实际的切片质量值与设定的切片质量值相比,其平均绝对误差为8.62 g,平均相对误差为7.41%,达标率为85.67%,该研究为猪大排的定量切片提供了一种有效方法。This study aims to improve the slicing accuracy and substantial material savings in frozen pork ribs.Three-dimensional reconstruction was also proposed for the quantitative program slicing on porcine large chops.Both the precision and efficiency of the slicing were optimized to enhance resource utilization in the meat processing industry.Advanced technologies were selected to revolutionize traditional meat slicing,ultimately contributing to operational efficiency and cost savings.Several key stages were utilized for the accurate and efficient slicing of the pork ribs.Initially,an experimental platform was developed for the point cloud acquisition.A line laser scanner was employed to capture the point cloud data of frozen pork ribs from multiple angles.The raw data was acquired to preprocess the point cloud.Data collection was realized to extract and isolate the target point clouds necessary for subsequent analysis.The data filtered out the noise and irrelevant information for accurate three-dimensional reconstruction.A registration was then conducted to accurately align the data because the point clouds were acquired from different viewpoints after the preprocessing.A rectangular target block was also utilized to accomplish the alignment in conjunction with the Trimmed Iterative Closest Point(ICP).The rectangular target block then acted as a reference point to align the various point clouds,while the Trimmed ICP algorithm facilitated the precise registration of the data.As such,all point clouds corresponded accurately to the actual physical structure of the pork ribs.Accurate alignment was essential to create a reliable three-dimensional model.The volume of the frozen pork ribs was then estimated after alignment.Graham scan algorithm was used to efficiently compute the convex hull of the point cloud,in order to determine the outer boundaries of the objects.The body slicing was segmented and then accumulated to accurately estimate the volume of irregularly shaped objects,like pork ribs.As such,the precise volume e

关 键 词：猪大排 点云数据 三维重建 体积计算 定量切片 

分 类 号：TS251[轻工技术与工程—农产品加工及贮藏工程]