Image Recognition On Special Pathogen Free Chicken Embryo

In recent years,a variety of avian influenza virus break out in the world,which result in a wide range of livestock's death and influence our human's health and life safety.SPF(Special Pathogen Free)chicken embryo plays an important role in the preparation of influenza vaccine,influenza vaccine is also related to the safety of human and livestock.At present,the mainly way to detect and classify chicken embryo is artificial methods in our country.However,this method has some disadvantages,for instance,low efficiency,large errors,low accuracy,and requiring a lot of labor which is easy to be affected by subjective factors.In this article,image processing and machine vision technology are used to detecting and classifying chicken embryo images.The specific contents are as follows:First of all,we need to detect to polluted embryo and unfertilized embryo.Because these two types of chicken embryo are different from other chicken embryo in appearance,polluted embryo and unfertilized embryo present obvious different color features under the same conditions.According to the difference of RGB color components on embryo image,we study the comparison of the mean and variance of chicken embryo RGB color components to distinguish the polluted embryos and unfertilized embryo.Secondly,we need to solve the problems of different kinds of irregular spot.Due to different physiological features and different light transmittance in chicken embryo,the images we collected exist different kinds of irregular spot.We need to study out an appropriate image processing algorithm to detect and filter those irregular spot.At first,extracting the ROI image area of chicken embryo,then detecting irregular spot by the Harris algorithm and spot detection algorithm based on RGB color components.Finally,comparing the results of two methods.Then,we need to solve the problem of classification of dead embryo,weak embryo and alive embryo.Due to dead embryo,weak embryo and alive embryo present obvious differences in the blood characteristics,we could identify three kind of embryos by the number of blood vessels and the area of blood vessels.First,grayscaling and denoising the image,then using the method of local adaptive WFCM algorithm to separate image features.Second,using morphological method to process the image.Finally,according to the number of blood vessels,embryo length,size and other characteristics of these three kinds of embryo to judge and identify chicken embryo.Finally,we develop a software system for the recognition and classification of chicken embryo images by using C# language,develop the image processing module and related programs to realize the online classification of chicken embryo images.