Vision-based Egg Defect Detection

With the rapid development of aquaculture in china,egg production and sales occupy a large number of international market,it is very important to egg quality testing.Egg surface has a lot of microbes and bacteria,in the process of egg transport,it will be caused by external factors such as knock against,extrusion to egg shell broken.Microorganisms and bacteria pollute easily the nutrients inside the egg and infect surrounding intact eggs,not only bring huge economic losses,but also harmful to consumers' health.It is necessary to develop detection technology of egg in the egg processing industry.Image detection technology based on vision technology has the high precision,high efficiency and no damage of advantages has been more attention in the egg processing industry.Shape defect detection commonly based on curve fitting method,curve fitting method often fitting a string line,circle and oval,but few of the egg shape detection research,the reason is that it is too difficult to accurate describe egg shape curve.Common curve fitting algorithm based on Hough transform and least square algorithm are greatly influenced by noise,the operation efficiency and fitting precision are low.Crack defects of the traditional detection algorithm based on texture feature and morphological algorithm,both have the problem how to save the maximum crack characteristics.So,the egg defects detection has important practical significance.This paper set up egg defect detection system based on vision,compare various light source parameters to choose the laser illuminator,compare camera properties to determine the CCD camera.Egg shape curve equation and the fitting curve of egg shape conditions were determined,using single mode and multimode RANSAC algorithm to fitting of egg shape,detection whether there is appearance defects.The high frequency enhancement filter is used to preprocess the image,then texture characteristics of the local biggest difference method and the combination of morphological processing to extract the crack characteristics,extract statistical crack pixel points,detection whether there is crack defects.According to the result of experiment,the proposed egg shape defect detection algorithm has robustness,and the accuracy is 9 5.4%,the recall rate is 96.1%;Egg crack defects detection algorithm accuracy is 96.8%,the recall is 97.5%.