| In recent years,many factory production lines have begun to use computer images to intelligently process efficient production and process simplification.Intelligence is a must for the transformation of every smart factory,which has obviously become an inevitable trend.Early studies have shown that the automatic detection device based on pork backfat has gradually changed from a hardware control equipment system to an automatic software and hardware combination system.By designing and implementing an intelligent detection and analysis platform,computer vision technology is used to measure the size of items.In the identification of inferior items and the intelligent classification of items,it is necessary to reduce the accidental error caused by manual labor and the cost of machine use in the factory.In order to explore the possibility that computer vision technology can detect and evaluate online meat quality,through the comprehensive analysis of the core indicators of pork backfat thickness and other auxiliary grade indicators,this paper mainly designs and realizes the intelligent detection of meat quality based on image processing.platform.In this paper,based on traditional image processing and computer vision related technologies,according to the analysis of the characteristics of the collected image data,according to the demand for real-time detection of pork backfat thickness and the task of how to achieve intelligent classification,this paper studies the detection method of pork backfat thickness.algorithm.First,this paper designs a series of image processing tasks for images with complex backgrounds,including bilateral filtering and denoising,binarization,morphological transformation,improved fuzzy C-means clustering algorithm for segmenting complex backgrounds,and Enhance its contour information based on the improved Canny edge detection algorithm.Second,in order to reduce the error caused by the complex backfat image laterally,an image correction algorithm based on multiple vanishing points is designed.Thirdly,the straight line fitting design is carried out to the contour information of the target image tortuous,which solves the problem of unsmoothness caused by the longitudinal direction of the image contour.The test results prove that the allowable error of this method is within 1mm of the upper and lower back fat,and it can be considered that the detection is correct.The accuracy rate of the final backfat thickness detection is as high as 95%,which meets the production requirements of the industry.Among them,the average time for testing a single product is about 0.5 seconds,which meets the real-time requirements of intelligent production.Based on the collection,storage,analysis and processing of data,the platform can detect the backfat of pork on the production line in real time and process the classification task of pork quality in time,and build an intelligent quality inspection platform based on image processing.This paper analyzes the needs of users and makes a general design.The platform architecture is divided into data storage layer,data processing layer,algorithm model layer,data transmission layer,business service layer and client.In terms of functional design,the business realization of login module,algorithm module,grading and classification module,intelligent analysis module and basic management module has been completed.Finally,after systematic testing,the meat quality intelligent quality inspection platform can meet the functional needs and usage needs of users,which can make the factory have a better direction of intelligent development,and also bring a lot of convenience to the user’s experience and use. |
