Research On The Detection Technology Of The Whole Surface Defects Of Jujubes

Jujube defect detection is an important link in the deep processing of jujube.With the rapid development of machine vision and automatic control in modern times,the artificial jujube detection system has been basically got rid of,and the automatic detection system based on machine vision and automatic control has become more and more mature.However,the currently used jujube defect detection equipment and technology still have disadvantages such as large size,low efficiency,and imperfect classification algorithm.Based on the machine vision and pattern recognition theory,this paper conducts an in-depth study on the key technologies of jujube full-surface defect detection.The main research contents are as follows:In view of the external defects of red dates,the functional requirements of the whole surface defect detection platform of red dates are determined;an efficient and stable whole surface image acquisition scheme of red dates based on optical reflection is designed;according to the conditions required by the acquisition scheme,the camera of the acquisition module is configured And lens selection,image acquisition and transmission methods,light source configuration methods,and platform construction.According to the interference on the image,the automatic brightness correction algorithm based on the Lambert reflection principle is studied,and the brightness of the image is uniformly adjusted;based on the BLOB analysis method,the target red date is extracted,and the adjusted red date image is background separated to get the background removed Target jujube image.Aiming at the defects of dried red dates which are rarely studied in the existing literature,the detection technology of red dates defects based on PCA-SVM is studied.The color moments and gray-level co-occurrence matrix are used to extract14-dimensional eigenvectors in the color and texture features of jujube.The principal component analysis algorithm is used to optimize the eigenvectors,and four main factor eigenvectors are obtained as the input of support vector machine.The crossover algorithm is used to determine the optimal support vector machine penalty parameter c and the kernel function parameter g to train the support vector machine multi-classification model,and use the trained model to perform multi-classification experiments on red dates.Experimental results prove that the model can classify red jujube defects with an accuracy rate of 97%,and the classification efficiency is high.Finally,the software system of the red jujube whole surface defect detection system is designed,and C# and Halcon are determined as the software development platform of this system.Designed the program interface of the jujube full-surface defect detection system,and designed the parameter setting interface.