Detection Of External Defects In Eggs Based On Machine Vision

Eggs with dirt stains or cracks must be removed in time so as to avoid cross infection.At present,most of China's egg processing enterprises mainly adopt manual inspection to pick out defective eggs with low working efficiency and high labor intensity.Study on the automatic detection of external defects of eggs will not only improve egg quality,but also improve the market competitiveness of egg processing enterprises.On the basis of previous studies,related theories and methods of image processing technology were used to detect external defects of eggshell in this paper.The prcesses such as image acquisition,preprocessing,stains and cracks recognition were carryed out on the laboratory experiment platform.And a series of experiments were designed to verify the proposed detection algorithm.The main research contents and achievements can be summarised as follows:(1)The egg image acquisition platform was designed to acquire the full details of the eggshell.In addition,to solve the problem that the black stains on the egg surface may cause the false division,a method combined with maximum contour extraction and seed filling algorithm was proposed in this paper.(2)Two eggshell stains enhancement methods respectively based on local texture and fast median filter were proposed to overcome the difficulty of detecting stains by traditional threshold segmentation,which was caused by the uneven illumination of the ellipsoid liked egg surface and complicated external stains.The experimental results showed that the recognition rate of dirty eggs was 98.4% and the recognition rate of clean eggs was 91.8%.(3)The negative LOG operator was adopted to enhance the cracks on eggshell under the condition of backlight transmission.Such method can not only overcome the interference of highlighted dark spots but also solve the problem that micro-cracks were not obvious on the egg image.The Hysteresis thresholding was adopted to acquire the ‘optimal' binary cracks image.Through the analysis of shape features of cracks region and thin spots region,the improved LFI index was adopted to distinguish the crack region from the mislabeled region.The experimental results showed that the recognition rate of cracked eggs was 92.5% and the recognition rate of intact eggs was 90.0%.(4)According to the functional requirements,the software programming was completed in the Visual Studio 2013 integrated development environment,using OpenCV and MySQL database.The detection system provided user management module,image acquisition module,defects detection module,display and inquiry of detection results module.The results of the above studies showed that the proposed external defects detection method can overcome some of the difficulties in the visual inspection of external defects in poultry eggs.The related detection algorithm has important significance for the realization of online automatic grading equipment for poultry eggs.