结合量子技术和FCM的猪肉颜色分级研究  

Study of Pork Color Grading Based on Quantum Technology and FCM

作　　者：孙宝 杨世恩[2] SUN Bao;YANG Shi′en(School of Computer Science and Technology, Southwest University of Science and Technology,Mianyang 621010, Sichuan, China;National University Science and Technology Park,Southwest University of Science and Technology, Mianyang 621010, Sichuan, China)

机构地区：[1]西南科技大学计算机科学与技术学院,四川绵阳621010 [2]西南科技大学国家大学科技园,四川绵阳621010

出　　处：《西南科技大学学报》2018年第4期83-89,共7页Journal of Southwest University of Science and Technology

基　　金：绵阳市涪城区创新创业重点项目(16zh0197)

摘　　要：猪肉颜色是猪肉分级评定时的一个重要指标。针对传统中值滤波存在的问题用量子理论的思想对其进行改进,改进算法不仅有较强的去噪效果而且能自适应判断模板内的像素点要不要进行处理从而能较好地保留图像细节信息。利用改进算法结合模糊C聚类(FCM)对猪肉图像进行处理分割,最后提取相应的颜色特征参数,基于这些特征参数运用多元线性逐步回归分析法建立了颜色等级预测模型。在样本测试实验中,所用方法对猪肉颜色等级的正确判定率达到94. 25%,在大幅降低算法复杂度的同时,准确率和鲁棒性都超过了其它同类算法。:Pork color is a significant characteristic of pork grading.In this paper,the improved median filtering algorithm not only has a strong denoising effect but also better preserves image detail information.With the improved modified adaptive median filter,corrective treatment of background color were used to keep original pork image,then the background of RGB images was removed by the method of Maximum Classes of red component,the Fuzzy C-Means Clustering algorithm was used to segment the muscle and fat,finally extract the color features of pork.The prediction model of color grade was established by multiple linear stepwise regression analysis.The sample set was used to identify the color grade,and the correct rate was94.25%.What's more,the accuracy and robustness of the algorithm are superior to other similar algorithms when the algorithm complexity is reduced.This method can provide a reference for study and application of online nondestructive detection of pork quality.

关 键 词：机器视觉 猪肉颜色分级 RGB颜色特征 肌肉分割 FCM 中值滤波 线性回归 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]