Research On Multiple Moving Fish Target Detection Algorithm For Aquaculture Monitoring

Video analysis has become an important technical means to promote aquaculture informatization and intelligentialize,moving target detection is the key link of fish tracking,behavior recognition,and abnormal behavior detection.In this paper,the problem of moving object detection in video surveillance of indoor aquaculture is discussed,and studying on moving object detection based on background modeling.The basic principle of the GMM method and the local binary patterns background modeling method was studied,including the single gauss model,mixture gauss model,LBP and its derivative operator.According to the actual situation of monocular indoor global aquaculture video monitoring,This paper discusses the challenges to the moving object detection algorithm,such as surface clutter,reflection and illumination change.Comparative experimental study was carried out for background modeling algorithm of aquaculture monitoring scene.The applicability of the GMM model and background model based on LBP is investigated,detailed analyzed the impact of parameters of each model to mixed gauss background modeling and the background modeling method based on LBP and its derivative operator.At the same time,the effects of different color space components are considered.The fish breeding surveillance standard data sets was built,using for objective evaluation of experimental results.Aiming at the problems of the lack of the target's integrity,ghost and light reflection noise,the original XCS-LBP texture descriptor was improved.Furthermore,a new target detection algorithm is proposed,which can adaptively adjust the similarity threshold of the local noise region and then separate the background and foreground.Finally,the improved background modeling algorithm is compared with the GMM algorithm and the original XCS-LBP algorithm,and the experimental results are analyzed qualitatively and quantitatively.The results show that the proposed algorithm is feasible and effective,can significantly inhibit the effect of "ghost" and the strong light reflection noise,used to solve the problem of moving object detection in indoor fish breeding monitoring scenes.