基于计算机图像分析的肌内脂肪含量测定

开发了一种快捷、准确的方法对肌内脂肪含量进行了测定。在Matlab操作平台下，应用计算机图像分析方法对大理石花纹含量的特征进行了提取；对与肌内脂肪含量相关性较强的理化指标，如：固体电导率、剪切力值、肌内干物质、灰分等进行了研究。并应用多元线性回归、非线性回归和神经网络等三种不同的数学方法，对肌内脂肪含量进行计算测定。其中，非线性回归模型正确率达到85％以上。

Mensurating Intramuscular Fat Content Based on Computer Image Analysis

An effective and exact method was developed to evaluate intramuscular fat content. Marbling content feature wasgathered by image analysis with Matlab. Some physical and chemical indexes such as solid conductance rate, shear force, dry matter,ash etc were studied. The research showed that the indexes had high correlationship with intramuscular fat content. Those indexeswere used to predict intramuscular fat content by 3 different mathematical models, such as poly-linearity regression, non-linearity regression and artificial neural network. Among those 3 models, non-linearity regression model had highest veracity andits correct rate is above 85%.