Research On Typical Product Feature Extraction And Classification Algorithm Based On Machine Vision

In recent years,with the development of international modern manufacturing industry,China's traditional industries are also being transformed into the intelligent manufacturing industry,in which machine vision technology has been rapidly developed and widely used because of its good performance.Machine vision has been widely used in the detection of agricultural products with its non-contact,high precision,fast speed,large amount of information,real-time,online and so on.It has greatly improved the quality and efficiency of agricultural products.Based on the machine vision technology,this paper studies the feature extraction and detection grading methods of two typical agricultural products(citrus and crab),and achieve the purpose of the study design and corresponding detection system using LabVIEW programming software.Firstly,two kinds of image preprocessing methods for two kinds of agricultural products are studied.Although the two processes are similar,including the image of the gray,image enhancement,edge extraction and morphological processing,but the appearance of the two products vary widely,after processing can highlight their own different characteristics.So the two products used by the algorithm to the type,order and parameters are very different.Then,the method of feature extraction for two kinds of products is studied.In terms of citrus,the size,maturity,fruit shape and defect area of the product were extracted by area method,green pixel occupancy ratio method and roundness method;Crabs:the use of area method,image recognition,color matching and other methods to extract the size of the product,male and female,fat and other characteristics.Finally,the product is graded according to the standard.After the product feature is extracted,the feature quantity information is converted into digital information,and the program is programmed according to the classification standard.And finally,the automatic grading of the product is realized.By experimental verification,the visual inspection system designed in this paper can identify the characteristics of the product more quickly and accurately,and has certain practicality.