In modern life, people have higher and higher requirements for material standards, high quality beef is sought by more and more diners, marbling beef is the best cow product in the world. Marbling density is a part of one of the important indices for evaluation quality of marbing beef, for these indicators usually adopt sensory evaluation, need professional evaluation personnel, it takes time and effort to train these professionals in the long run, Therefore, based on this situation, the development of a new type of computer evaluation system has a good prospect and research value. The main content of this study is based on the basic principle of RGB-HSV color model. The RGB values of the measured samples are then converted into HSV values for data processing to observe whether there is a rule in the data and summarize its linear equation. The results showed that different grades of beef had significant differences in RGB-HSV color model analysis, the linear relationship is, y(marbling standard)=0.018x( V value)+0.664 $(\mathbf{R^{2}}=\boldsymbol{0.96533})$), with this data model, we checked our database for the marbling beef, the average accuracy is about 98.76, which also proved the effectiveness of the developed computer evaluation system and the feasibility on the method.