Research On The Estimation Model Of Scophthalmus Maximus' Weight Based On Depth Image

Fish's body weight measurement is one of the important foundations in health monitoring in fisheries aquaculture.Scophthalmus maximus generally have the characteristics of large area,thin body,and swimming slowly.Traditional Scophthalmus maximus'body weight uses manual observation and measurement,These measurements not only spending more time and laborious,but also subject to factors that make the accuracy unstable.Computer vision method based on image processing has the advantages of time saving,intelligence,automation and so on.Which become hotspot in the field of this research in modern fishery monitoring.In this paper,the image monitoring and body weight estimation were carried out with the Scophthalmus maximus as the research object.The main research contents are as follows:(1)Reseach depth image data extraction,image preprocessing and image segmentation methods.The original depth image of turbot was preprocessed by means of median filtering,Otsu threshold segmentation and contour extraction.The pre-processed images are used to extract the four feature vectors of the body length,body width,perimeter and area of the turbot.At the same time,the depth of field data in the image is extracted to provide data support for mapping modeling.(2)Establish growth data and depth information mapping model.The depth information expresses the target and the position of the camera in the depth image.The experiment uses the infrared ray of the Kinect camera to obtain the depth information in the image of the Scophthalmus maximus as the independent variable,and the actual body length of the Scophthalmus maximus as the dependent variable,the body in the image.Long,data fitting is performed under matlab,and the fitting result is applied to the growth data-mapping model.The decision coefficient of the mapping model is R~2=0.9236,RMSE=0.000632cm.That can be seen that based on the fitting algorithm of this paper,the fit between the depth and the growth data of Scophthalmus maximus is better.(3)Research on the weight estimation algorithm of Scophthalmus maximus based on grid algorithm optimization support vector regression.We get the actual growth data through the growth data and depth information mapping model,and the feature parameters are normalized and composed into feature vector groups.As the input vector,the grid-based algorithm is used to optimize the support vector regression for training modeling and mass quality.Estimate.According to the characteristics of the phenotypic characteristics of Scophthalmus maximus,the four characteristics of body length,body width,perimeter and area of Scophthalmus maximus were selected as the characteristic parameters of body mass estimation,and the mass estimation model of Scophthalmus maximus was constructed.The results show that the estimated results are compared with the actual measured results.RMSE=0.0297g,R~2=0.9901.The model has the characteristics of simple and flexible,high estimation accuracy and easy to implement,and has a good application prospect.(4)Using Ecplise.Matlab and Microsoft Visual Studio 2017 software to develop the turbos quality estimation platform,the system includes three parts:image acquisition,model construction and weight estimation.The image acquisition module mainly uses the video data collected by the camera to intercept the Scophthalmus maximus image.The model is constructed to perform feature vector input and model training on the acquired image using the constructed classification model.Body weight estimation Based on the trained model and input vector,the mass of the Scophthalmus maximus was estimated and stored in the server to construct the Scophthalmus maximus growth database.In this paper,the depth image of the Scophthalmus maximus is used as the experimental material,and the growth data-mapping model is used for correction.Based on the grid optimization,the support vector machine algorithm is used to estimate the eigenvector.From the results,the body mass estimation model can effectively estimate the body mass of Scophthalmus maximus,which can provide support for the monitoring of Scophthalmus maximus and provide an important basis for establishing the growth database of Scophthalmus maximus.