Beef Grading Method Basing On The Marbling Characteristics

Beef quality grading is significant in the economic benefit.Comparing to no grading beef,the price of hierarchical beef is several times higher, even ten times.Beef marbling level is the most important indicators in the evaluation of beef quality.The fractal dimension and marbling area ratio are closely related to marbling level. The assessment of beef quality, marbling quality of eye flesh is the primary indicator.Eye flesh is cross section of the longissimus dorsi. Intramuscular fat in longissimus dorsi of cow carcass,which is between12-13or6-7sternal rib,is called marbling.All countries in the current world,beef grading method mainly is visual assessment. Determining the marbling grade is on the basis of experience by professional beef quality rating personnel, that to observe abundance of intramuscular fat in longissimus dorsi of cow carcass,which is between12-13or6-7sternal rib,then to compare with the standard of beef marbling chart in each countries.It is subjective with this approach.This topic absorbs some of forefathers’ research results about beef grading.Prediction mathematics model of beef marbling grade is established.This paper mainly completed the following several aspects.1)Standard chart of beef marbling in China and the United States, extraction method of their longissimus dorsi and the fractal dimension calculation method are studied. Methods of extracting longissimus dorsi is found, and longissimus dorsi is extracted step by step. According to some special images that longissimus dorsi compact connect proud flesh, using artificial adding ellipse method, iterative corrosion and expansion method, artificial adding curve method. Through comparison, the best way in these three methods is to manually add curve method. The calculation of fractal dimension use the traditional box-counting dimension method, differential box-counting dimension, information dimension method, improved box-counting dimension method. Four methods are compared, and the best method is improved box-counting dimension method.2)60beef eye muscle samples are collected.Pictures of the sample are photon,and samples are carried on the classification. Matlab software is step by step used to extract longissimus dorsi with several procedures.Some samples, whose eye flesh and longissimus dorsi connect closely, several methods are compared, and the best way is to manually add curve method.3)Two characteristic parameters of marbling,which are the fractal dimension values and marbling area ratio,are determined.Single factor analysis is used, and influence on level of the two characteristic parameters is extremely significant. 4)The forecast model of the marbling grade is established by basing on BP neural network.Two samples are selected randomly in level3,4,5,6,7.The datas of50samples are used to predict level of other10samples.The accuracy is73%.Throughout both at home and abroad, the current level evaluation of beef is mainly adopt to the method of artificial visual.This method is a little subjective. The knowledge of machine vision,BP neural network etal are used in beef grade,and the effect is well.Machine vision, fractal theory and computer image processing technology are the most promising technology in the study of achieving beef grading.lt is considered by meat scientists all over the world. All these methods are reflected in this thesis.