基于HOG和GA-SVM的鱼类疾病图像识别算法的研究  

Research on Fish Disease Image Recognition Algorithm Based on HOG and GA-SVM

作　　者：贾童 JIA Tong(Taiyuan Normal University,Jinzhong 030619,China)

机构地区：[1]太原师范学院,山西晋中030619

出　　处：《电脑与电信》2024年第12期17-21,共5页Computer & Telecommunication

摘　　要：渔业一直是我国农业的支柱产业之一,但因为很多从业者专业能力不足,导致鱼的养成率一直不高,造成的经济损失巨大。针对传统渔业养殖存在的智能化水平低、效率低等问题,提出一种基于HOG和GA-SVM支持向量机的鱼类疾病的图像识别方法。本研究以鱼类疾病图像作为实验对象,采用HOG特征描述符提取图像特征信息,采用线性判别分析算法LDA进行降维处理,将降维后的新向量作为输入向量代入SVM分类器进行训练。模型在经过训练后,对于鱼类疾病的识别率达到92.473%,对于研究机器学习在渔业上的应用提供一定参考。Fisheries has always been one of the pillar industries of agriculture in China,but due to the insufficient professional abilities of many practitioners,the cultivation rate of fish has been low,resulting in huge economic losses.This paper proposes an image recognition method for fish diseases based on HOG and GA-SVM support vector machine to address the problems of low intelligence level and low efficiency in traditional fishery aquaculture.This study takes fish disease images as the experimental object,extracts image feature information using HOG feature descriptors,and uses linear discriminant analysis algorithm LDA for dimensionality reduction.The reduced new vector is used as the input vector to train the SVM classifier.After training,the model achieves a recognition rate of 92.473%for fish diseases,providing some reference for studying the application of machine learning in fisheries.

关 键 词：渔业养殖 HOG GA-SVM 机器学习 图像识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]