Application Of Computer Vision In The Automatic Grading For Beef Marbling

Beef is one of the most popular food, people more and more like it because of its high-protein, low-fat, delicious taste , and so on. Today, the total output of beef in P. R. China is about 20 times of 1980's after rapidly development in past 20 years. But it has been difficult for beef of our country to export to developed country. The reason our country can't play an important part in international beef market is that most of them is non-grade commodity. To accord with the development of international market, our country had constituted the beef grading system.The beef marbling grade is the dominant parameter in deciding the meat quality. Traditionally grading of beef marbling has been performed by human graders. However, the sensory inspection is subjective, time-consuming, labour-intensive and the accuracy of the test can not be guaranteed. By contrast Computer vision has great potential for beef grade objective evaluation. Image processing techniques can quantitatively and consistently characterize complex color, geometric and textural properties. So this research is to study an auto-inspecting method of beef quality grades using computer vision based on marbling according to our country's standard. The work that done and accomplishment are as follows:(1) The hardware of computer vision for beef carcass grading was designed. Diffusion reflection light boxes were used to capture clear image in order to eliminate blaze produced by reflection of water and oil.(2) The preprocess algorithm was designed. The background of sample images was removed by the Method of Maximum Classes Square Error (OSTU) in the red color band of the RGB image and noise was eliminated by median filter.(3) Longissimus dorsi (l.d.) was segmented from ribeye imaging based on Ohta & RGB color systems. Marbling was extracted from l.d.(4) Using a fuzzy C means (FCM) clustering algorithm.(5) The feature of beef marbling was extracted based on area and moment.(6) Applying Support Vector Machine(SVM) classification, the forecast precision was 84.9%.(7) The software of computer vision for beef carcass grading was designed.