基于 K-means-RBF 多源数据融合的鸡肉品质分类方法研究

摘要：为了实现肉鸡贮运过程中鸡肉品质的快速无损检测，利用高光谱成像技术的特点，对图像进行图谱特征和光谱组 合特征提取，提取反映鸡肉内在品质的光谱特征；同时提取反映鸡肉外在特征的图像纹理特征、颜色特征，采用多源数 据融合方法建立鸡肉品质快速分类的 K-means-RBF 预测模型。结果表明，采用 K-means-RBF 神经网络建立的多源数据 融合检测模型比单一特征分类模型具有更高的预测精度，精度可达100%，证明了多源数据融合方法在鸡肉品质检测中的 有效性和必要性。

关键词：鸡肉；高光谱；多源数据融合；神经网络

期刊：Chinese Control and Decision Conference, CCDC

DOI：10.1109/CCDC.2018.8407167

设备：GaiaField

Study on Chicken Quality Classification Method Based on K-means-RBF Multi- source Data Fusion

Abstract：In order to realize the rapid and non-destructive testing of chicken quality in bDOIler storage and transportation, the characteristics of hyperspectral imaging technology was used for the map and spectrum combo features, to extract the spectral characteristics, which that reflect the intrinsic quality of chicken; and the image texture features, color features, which reflect the external characteristics of chicken were also extracted. The multi-source data fusion method was used to establish K-means-RBF prediction model of chicken meat quality rapid classification. The results show that the multi-source data fusion detection model established by K-means-RBF neural network has higher prediction accuracy than single feature classification model, and the precision is up to 100%. It is proved that the validity and necessity of multi - source data fusion method in chicken quality testing.

Keywords：chicken; hyperspectral; multi-source data fusion; neural network