The Grading Assessment Of Tan Sheep Carcass Based On Computer Vision Technology

Establishing mutton grading system is an important measure which can guide the market consumption trends, regulate mutton circulation channels and promote the healthy development of the mutton industry. At present, the grading assessment of sheep carcass still remains on the level of manual operation due to technical reasons. In the process of manual grading, there are many factors which can greatly influence the grading results, such as the graders’ sensory differences, their experience and working fatigue.Methods of traditional sensory evaluation and chemical manual detection are difficult to meet the requirement of quick online detection. Therefore, this paper takes Tan sheep meat, Ningxia’s characteristics agricultural products, as the studies object, and puts forward the grading assessment of mutton yield and its quality based on computer vision technology by means of image information. It provides a theoretical basis for the automatic grading assessment of mutton, which will replace the manual evaluation method.The main conclusions are as follows:(1) In this paper, industrial video camera is used to collect high-quality rib-eye section images of sheep carcass. The processing method of rib-eye section images is mainly studied. First, rib-eye section images are pre-processed based on gray processing 3x3 median filtering and Otsu algorithm. Then, the image binaryzation is carried on and by means of regional stepwise segmentation does get rough extraction of the rib-eye region and the effective rib-eye area is accurately extracted by means of regional labeling and morphological processing. Finally, the eigenvalue of rib-eye color and marbling area are detected.(2)The software spss20.0 is used for correlation analysis of four factor:carcass meat yield, carcass weight, rib-eye area and back fat thickness. The results show that correlation coefficient of carcass meat yield and back fat thickness is the largest, reaching 0.919; the correlation coefficient of rib-eye area and carcass meat yield reaches 0.706; the correlation coefficient between them is more than 0.706, which means that their correlation is high. Through multiple linear stepwise regression analysis does get the quota of carcass yield grading:rib-eye area and back fat thickness. Finally, the prediction equations of meat yield is established, which is interacted by these two factors. When the test set samples are used to identify the yield grading, the correct judgment rates of the model is 90%.(3) The color’s features of the rib-eye mainly consist of mean value and standard deviation of R, G, B, H, S, I, which are extracted by using MATLAB software. And the marbling’s features include ratio of total fat area, ratio of big fat particle area, ratio of small fat particle area, total number of fat particles, numbers of big fat particles and small fat particles. By means of multiple linear stepwise regression analysis method does respectively establish the prediction model of color and marbling grading. When the test set samples are used to identify the grading, the correct judgment rates of color grading and marbling grading are 90% and 85% respectively.