Study Of The Application Of Computer Vision Technology Used In Beef Tenderness Prediction

Beef tenderness is an important index to evaluate eating quality of beef. The two major methods to evaluate the beef tenderness are visual evaluation and shear force determination. The method of visual evaluation is always influenced by objective factors and there is always large error; the result of shear force determination is much exact, but the process is complicated and time-consuming which is a destroy test. This particle combines the machine vision technology with a priori knowledge of human and takes the beef rib-eye image as the study object to collect, process, analyze and identify the pictures to complete the study of beef tenderness prediction.The main content and result of this study are as below:1. By analyzing the rib-eye of beef image, it establishes suitable the machine vision system in order to reduce the influence of the beef surface’s light reflection and roughness in image acquisition. The system is combined by beef tenderness prediction software system and hardware system forming of light module, image acquisition module and computer system which ensures the accuracy of tenderness.2. Apply image processing methods for the collected beef rib-eye image to do image preprocessing, area segmentation and collect from the effective rib-eye area and connective tissue in this area, during the study of fascia color in beef rib-eye, it raises a study for fascia recognition segmentation method in colorful beef image which is from improved fuzzy c-means clustering algorithm, which optimize the initial cluster centers selection and the distance weighted formula of FCM clustering algorithm. The test result shows that the average accuracy of this method is near98%for the fascia area segmentation in beef rib-eye image and the operating time of this method is only a half of that traditional FCM costs.3. In the study of beef tenderness forecast model establishment, it analyzes the relationship between the feature index and its tenderness, and confirm the feature index which is closely related the tenderness. It applies SPSS to collect and analyze all the test data and finally uses multiple linear regression analysis to establish the math model of beef tenderness prediction. It also verifies the prediction accuracy of beef tenderness model from the collected sample data whose prediction average error is less than0.3kg.It studies the relationship between the beef rib-eye image and its tenderness in a deep-going way in this particle, and designs a computer vision system which is much suitable to collect the beef tenderness feature index. This system lays good foundation of the image calculation method completion and Tenderness prediction model establishment. The method raised in this article to predict beef tenderness is much better after test.