Research On Nondestructive Testing And Automatic Grading Of Kiwifruit Based On Computer Vision

On-line detection and grading of kiwifruit based on computer vision is of great significance for improving the competitiveness of China's kiwifruit market.Not only does it improve the sorting efficiency and fruit quality,but also reduces the cost of grading.At the same time,automatic detection and grading technology also reduce the fruit surface.Damage,thereby increasing fruit sales and commodity value.At present,there are shortcomings such as low efficiency,high misjudgment rate,and slow speed in the process of kiwi fruit classification,and it is difficult to meet the requirements of real-time classification.Based on this,this paper proposes the research of kiwi detection algorithm based on computer vision.Kiwi fruit grading is an important part of the post-natal treatment of fruit.In view of the low sorting rate and grading rate of kiwifruit in China,which affects the current value of commodities,manual kiwifruit grading has the disadvantages of time-consuming,laborious,high cost,and poor effects.Computer vision technology grading has speed Fast,large amount of information,high accuracy and other advantages.By describing computer vision technology,the shape,size,surface defects,color and other characteristics of kiwifruit are detected,and then grading processing is carried out,which improves work efficiency and reduces costs.In order to improve the quality of image acquisition and reduce image processing time,the choice of background color of kiwifruit in image acquisition was studied.The study shows that a black background helps to separate the target from the background.To improve the image processing speed,the image is used in the underlying information processing of the image Grayscale,image smoothing,image segmentation and other preprocessing operations.In terms of kiwifruit shape and size classification,based on the analysis of traditional fruit feature extraction methods,the extraction of kiwifruit appearance size,shape features,area and volume and other features are proposed,in which Fourier descriptors and image invariant moments are used..When grading kiwifruit,the HSI model was used to describe its color characteristics,and based on the kiwifruit color characteristics,a BP neural network based on particle swarm optimization was established.The test results show that the grading accuracy is high and can meet the requirements of grading The classification speed is fast,and the accuracy rate is as high as 95.6%.The surface defects of kiwifruit were treated by morphology.The BP neural network algorithm based on genetic algorithm was used to classify the surface defects.The accuracy of the experiment was as high as 91.3%.Therefore,the kiwi fruit grading detection system designed in this paper has verified that the system meets the requirements of real-time grading through experiments,and also verifies the feasibility and correctness of the theoretical research and design scheme proposed in this paper.