基于深度图像的大菱鲆体重估测模型研究  

Research on the Estimation Model of Scophthalmus maximus’ Weight Based on Depth Image

作　　者：田洁 杨信廷[1,2] 徐大明[2] 孙传恒[2] 吝凯 周超[2] TIAN Jie;YANG Xinting;XU Daming;SUN Chuanheng;LIN Kai;ZHOU Chao(College of Information Technology, Shanghai Ocean University, Shanghai 20000;Key Laboratory of Information Technology, Ministry of Agriculture and Rural Affairs,Beijing Agricultural IOT Engineering Technology Research Center, National Agricultural Information Engineering Research Center, Beijing 100097, China)

机构地区：[1]上海海洋大学信息学院,上海201306 [2]国家农业信息化工程技术研究中心,农业农村部信息技术重点试验室,北京市农业物联网工程技术研究中心,北京100097

出　　处：《中国农业科技导报》2019年第9期90-96,共7页Journal of Agricultural Science and Technology

基　　金：国家重点研发计划项目(2017YFD0701705)资助

摘　　要：为了对大菱鲆体重进行无损估测,提出了一种基于深度图像的大菱鲆体重估测模型。该方法首先对大菱鲆深度图像进行图像预处理,提取出深度信息与生长数据进行映射建模,拟合出目标特征,并结合网格搜索优化支持向量回归(GS-SVR)算法,实现大菱鲆体重估测。估测结果与实际测量结果进行对比,两者的决定系数(R2)达到0.990 1,均方根误差(RMSE)为0.029 7。该模型具有简单灵活、估测精度高和易于实现等特点,同时具有很好的应用前景。In order to carry out non-destructive estimation of the weight of Scophthalmus maximus, a weight estimation model was proposed based on depth image. The method firstly performed image preprocessing on the depeth image, extracted the target feature by using the growth data-mapping model, then combined the grid search optimization support vector regression (GS-SVR) algorithm to realize the weight estimation of Scophthalmus maximus. The estimated results were compared with the actual measured results. Experiment showed the optimal fit (R2) was 0.990 1, The root mean square error (RMSE) was 0.029 7. The model had the characteristics of simple and flexible, high estimation accuracy and easy implementation,which has a good application prospect.

关 键 词：大菱鲆 深度图像 映射模型 体重估测 

分 类 号：S126[农业科学—农业基础科学] TP391.41[自动化与计算机技术—计算机应用技术]