基于机器视觉的鲫鱼和鲤鱼品种识别方法研究  

Study on crucian and cyprinoid species identification method based on machine vision

作　　者：万鹏[1] 潘海兵[1] 宗力[1] 陈红[1] 

机构地区：[1]华中农业大学工学院/国家大宗淡水鱼加工技术研发分中心(武汉),湖北武汉430070

出　　处：《广东农业科学》2012年第17期184-187,共4页Guangdong Agricultural Sciences

基　　金：国家自然科学基金(61007058);中央高校基本科研业务费专项资金(2010QC006);华中农业大学引进人才科研启动基金(5220409079)

摘　　要：淡水鱼品种的准确识别是淡水鱼深加工的前提。以鲫鱼和鲤鱼为对象,研制了鱼体品种识别装置,利用机器视觉技术对鱼体样本进行检测,获取鱼体样本图像;通过图像处理技术对鱼体图像进行分析处理,将鱼体分为5段,计算各段的平均宽度与各段长度的比值,获取鱼体形态特征参数,再以5个形态特征参数作为输入值,构建BP神经网络对鱼体品种进行识别。试验结果表明,该方法对150-500 g范围内的鲤鱼和鲫鱼识别准确率可分别达到100%和93.33%。The accurate identification of freshwater fish species was the premise of the deep processing of freshwater fish,this paper studied the identification method of fish species based on machine vision technology taken the crucian and cyprinoid as the object.The paper also developed the identification device of the fish species and detected the samples of the freshwater fish using the machine vision technology,got the pictures of the freshwater fish;then the pictures of the fish body were processed and analyzed using the image processing technology,and the fish bodies were divided into five segments evenly,then calculated the ratio of the average wide and the long of each segment,obtained the morphological characteristic parameters of the fish body,which were as input parameters,and constructed the BP neural network to identify the fish species.The test results showed that the accuracy of the methods identified the crucian and cyprinoid weighing with in 150-500 g could reach 100% and 93.33%,respectively.

关 键 词：食品机械 机器视觉 品种识别 鱼体 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]