Study On Eggs Quality Grading Detection Based On Machine Vision

In this paper,we mainly use artificial recognition method for egg detection in our country,which is easily affected by subjective judgment,work experience and personal emotion.We often use the combination of machine vision and computer image processing technology to study egg detection:(1)This paper analyzes the current research situation and existing problems of machine vision in egg recognition and detection at home and abroad,and designs a machine vision plus computer image processing system,which involves the selection of camera,lens,light source and irradiation mode.The camera selected in this paper is AVT face array CCD industrial camera,model AVT Mako G-125 B,lens is industrial lens of Computar company,model is zoom lens,image processing software configuration win10 system,plus image processing database.(3)For the detection of egg crack,the second order negative LOG edge detection operator is used to obtain better contour extraction effect,and it can effectively suppress the interference of the background region.By calculating the circularity,aspect ratio and narrow index of the typical crack area and interference area,it is found that the narrow index is used as the threshold to distinguish the interference area from the crack area,and the threshold T is 12.Using the crack detection algorithm mentioned in this chapter,the detection rate is 95% for the complete egg,97.5% for the crack egg,and 96.3% for the whole egg.(4)The yolk is processed in the RGB color space,and the yolk feature image is extracted.for the gas chamber,the color space conversion is used to convert the RGB into the HIS color space,and then the gas chamber feature image is extracted.In order to test the error of comparing the collected image with the original image,the correlation coefficient between the long axis short axis and the real long axis short axis of the image is more than 0.97 by drawing fitting.Compared with the image egg shape index and the image egg shape index,the maximum error is 0.013,indicating that the image collected is very close to the real value.The models of egg yolk area ratio and gas chamber area ratio and HAV value were established respectively,in which the gas chamber area ratio had strong correlation with the model established by HAV value,indicating that the gas chamber area could reflect the freshness of eggs more.It was found that the correct rate of the model was 93.3% with the egg yolk area ratio and 96.7% with the gas chamber area ratio.