基于多信息融合的鸡蛋新鲜度检测研究  

Detection of Egg Freshness Based on Multi-information Fusion Technology

作　　者：徐彦伟[1] 崔建鹏[1] 颉潭成[1] 南翔[1] 

机构地区：[1]河南科技大学机电工程学院,河南洛阳471003

出　　处：《中国家禽》2014年第8期32-36,共5页China Poultry

基　　金：国家自然科学基金项目(51305127);河南科技大学青年基金项目(2012QN024)

摘　　要：为了提高鸡蛋新鲜度检测的准确率和稳定性,应用多信息融合技术,通过BP神经网络将机器视觉和光照度两种传感器采集到的信息在特征层进行融合,构建了鸡蛋新鲜度融合模型。在模型的建立过程中,分别选取蛋黄与整蛋面积比值和鸡蛋透光度作为特征参数,建立这两个特征量与新鲜度的关系模型。经检验,通过该方法对鸡蛋新鲜度识别的准确率为92.5%。验证试验结果表明:基于机器视觉和光照度的多传感器融合技术检测鸡蛋新鲜度是可行的,检测结果的准确性和稳定性相对于单个模型有明显的提高。In order to improve the accuracy and stability of egg freshness detection, through application of multi-information fusion technology and BP neural network, machine vision and illumination information collected by the two sensors at the feature level were fused to construct egg freshness fusion model. In the process of establishing the model,yolk and whole egg area ratio and egg transmittance were selected as the characteristic parameter to establish the relational model for these two features quantity and freshness. Upon examination, egg freshness recognition accuracy of this method was 92.5%. Verification test results showd that through multi-information fusion,based on machine vision and illumination technology,multi-sensor fusion technology detecting egg freshness is feasible,the test results compared with the accuracy and stability of the single model has significantly improved.

关 键 词：多信息融合 鸡蛋 机器视觉 透光性检测 BP神经网络 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]