|
Application of machine vision technology in detection of egg quality at home and abroad, has been widely carried out, and is increasingly becoming one of the important means of detection. With the technology of machine vision and image processing technology to develop specialist application prospect of machine vision technology in the field of egg quality detection is becoming increasingly broad. Eggs quality testing external quality testing and internal quality testing of two parts, the size of the eggs, egg shape index, weight and contents, freshness egg external quality testing and internal quality inspection contents and index. This article based on machine vision technology, based on the edge egg size detection algorithm, linear regression models and other technical methods to detect egg size, egg shape index, weight-based external quality indicators and freshness main contents internal quality indicators, and use the SOM neural network classifiers were grading eggs by weight indicators. The main research contents and results of this study are as follows: 1, the use of machine vision technology at home and abroad egg external quality and internal quality detection research progress and the status quo, and pointed out that a similar study carried out based on Egg Quality detection methods of machine vision research and testing equipment development needs. 2, establish and perfect eggs suitable for the study of external and internal quality inspection machine vision systems. Egg external quality detection system consists of a light box the six fluorescent lamps (F40BX/840,), color CCD camera, image capture card, the ADVANTECH INDUSTRIAL COMPUTER 610 IPC; egg internal quality detection system consists of a light box, a frosted glass bulb incandescent (PHILIPS), color CCD camera, image capture card, ADVANTECH INDUSTRIAL COMPUTER 610 IPC. 3, collected egg external quality image, preprocessing the image, using the instruction value threshold value from a combination of R, G, B color components for image segmentation using Laplacian extracted eggs edge, use size detection algorithm based on the edge of the egg and a linear regression model detection eggs longitudinal diameter, maximum diameter, and egg shape index, longitudinal diameter, maximum diameter and egg shape index detector model correlation coefficient were 0.9923,0.9816 and 0.9579 . 4 egg size detection algorithm to extract eggs longitudinal diameter, maximum diameter, transverse diameter and under transverse diameter the four size amount, with the amount of four dimensions as independent variables, egg weight as the dependent variable, egg weight detection Multiple linear regression model, the model correlation coefficient of 0.9781, egg weight detection of the absolute error of ± 3 grams or less. SOM neural network classifier eggs according to their weight indicators divided into less than 55 grams, 55 grams to 65 grams (not including 65 grams), 65 grams and 65 grams or more of three levels, classification accuracy rate of 90.6%, 76.8% and 82.5%, respectively. 5, the acquisition of the transmission image of the egg contents, by preprocessing the transmission images, the extracted egg contents transmission target R, G, B, H, S, the mean value of the characteristic component of the I-six colors. Measured using an electronic balance egg weight, height vernier caliper measurement was egg albumen height, Haugh unit values ??calculated characterization egg freshness. Six color characteristics of the target component of the egg contents filtered transmission mean as the independent variable, egg Haugh unit value of the dependent variable, namely the establishment of the red shell eggs and white shell eggs freshness detect multiple linear regression model, model The correlation coefficient of 0.8674 and 0.8929, respectively.
|