A method and system for evaluating fattening pork meat quality based on machine vision

Abstract

The invention discloses a method and system for meat quality evaluation of fattened pork based on machine vision, which collects pork images of pork to be tested; divides the pork images into multiple sub-region images; and screens all muscle sub-images in each sub-region image as muscle sub-images. Image collection; establish a fast scan sequence table for each muscle sub-image in the muscle sub-image collection; sequentially detect whether there is a muscle sub-image whose brightness is higher than the acidification brightness threshold in each muscle sub-image of the muscle sub-image collection; if there is a muscle sub-image whose brightness is higher than the acidification brightness threshold The muscle sub-image of the brightness threshold marks the pork to be tested as stale pork, otherwise it marks the pork to be tested as fresh pork; forms a quantitative evaluation to quickly and accurately evaluate the quality of pork, and does not depend on a specific hardware device structure , the real-time detection effect on the pork image is good, the detection efficiency is high, the image processing speed is fast, and the result of the pork quality can be obtained quickly.

Description

technical field [0001] The disclosure belongs to the fields of machine vision processing and meat freshness quality inspection and dispatching, and in particular relates to a method and system for evaluating fattening pork meat quality based on machine vision. Background technique [0002] In terms of pork quality inspection, the current quality inspection of pork is mainly through manual subjective evaluation, and the evaluation is carried out according to the standard color map of pork. Because manual judgment is relatively subjective, and there are large errors in a large number of comparisons, it is impossible to form Quantitative standards cannot quickly and accurately evaluate the quality of pork. Not only is the process cumbersome, but the test results of different people are often different. A large part depends on the professional level of the evaluators, and it is difficult to objectively form a quantitative evaluation. [0003] In the existing technology of using ...

AI Technical Summary

Problems solved by technology

[0002] In terms of pork quality inspection, the current quality inspection of pork is mainly through manual subjective evaluation, and the evaluation is carried out according to the standard color map of pork. Because manual judgment is relatively subjective, and there are large errors in a large number of comparisons, it is impossible to form Quantitative standards cannot quickly and accurately evaluate the quality of pork. Not only is the process cumbersome, but the test results of different people are often different. A large part depends on the professional level of the evaluators, and it is difficult to objectively form a quantitative evaluation.

[0003] In the existing technology of using machine vision to detect the quality of livestock meat, the invention with the publication number KR100514471 (B1) uses near-infrared spectroscopy to detect the freshness of meat, and only establishes the relationship between spectral information and volatile basic nitrogen indicators. For predicting the freshness, the index is too single, and it is difficult to evaluate the freshness of meat; and the invention with the application number CN200910098332.7 discloses a real-time detection and grading method of pork appearance quality based on machine vision, by collecting pork images; Real-time image processing and feature extraction; obtain feature information reflecting the appearance quality of pork; then use the prediction model to evaluate the meat quality, complete the color, marbling, tenderness and comprehensive quality grade evaluation of the eye muscles, and complete the muscle color, Grading of obesity and comprehensive quality

Although it can detect and grade single quality indicators such as muscle color, marbling, and tenderness, as well as comprehensive quality, it relies on a specific hardware device structure, and requires the construction of complex mathematical models, and the time complexity and space complexity are too high. , it is difficult to achieve fast real-time detection effect, the detection efficiency is low, and the effect is not good

Images