基于LBP特征的BP神经网络鱼群摄食行为检测研究  

Study on the Detection of Fish Feeding Behavior Based on BP Neural Network of LBP Feature

作　　者：朱瑞金 张涛 扎西顿珠 Zhu Ruijin;Zhang Tao;Zhaxi Dunzhu(College of Electrical Engineering,Tibet College of Agriculture and Animal Husbandry,Linzhi 860000,China)

机构地区：[1]西藏农牧学院电气工程学院,西藏林芝860000

出　　处：《黑龙江科学》2021年第18期5-8,共4页Heilongjiang Science

基　　金：西藏自治区自然科学基金(XZ2019ZRG69)。

摘　　要：为精准判断鱼类摄食时间及规律,提升鱼群智能投喂技术水平,基于图像处理及人工智能技术,提出了一种基于LBP特征的BP神经网络模型的鱼群摄食行为的检测识别算法。利用对鱼群的摄食活动图像及未摄食活动的图像数据,分别进行滤波等预处理及LBP特征提取,将特征纹理数据送入BP神经网络模型进行训练,利用训练好的神经网络模型进行鱼类摄食行为的分类识别。经过实验仿真,由实验结果可得设计的算法对鱼类摄食行为的识别准确率可达97.23%,来实现精准投喂,促进水产养殖规模化发展。In order to accurately judge the ingestion and regulation of the fish,and improve the technical level of intelligent fish feeding,based on picture processing and artificial intelligent technology,the paper proposes the algorithm for the detection and recognition of fish feeding behavior based on the BP neural network model of LBP features.After preprocessing,such as filtering and LBP feature extraction on the feeding activity image of the fish feeding and the image data of the unfeeding activity respectively,the characteristic texture data is sent to the BP neural network model for training,and then the trained neural network model is used for classification and recognition of fish feeding behaviors.Through experimental simulation,it can be obtained from the experimental results that the algorithm designed in this paper can achieve 97.23%accuracy in identifying fish feeding behavior.

关 键 词：机器视觉 LBP特征 局部纹理 BP神经网络 

分 类 号：TP36[自动化与计算机技术—计算机系统结构]