| China is a large citrus producing country,the speed and accuracy of citrus sorting directly affects the competitiveness of China’s citrus in the international market.At present,China’s citrus sorting mainly relies on manual completion,the method not only consumes a lot of labor resources and grading precision is not high,grading efficiency is low.Nowadays,although there are some fruit non-destructive testing grading production lines,but most of these production lines are large and expensive,and most of the equipment with detection of internal or physical and chemical quality rely on near infrared,hyperspectral technology,making it more expensive and difficult to maintain.Therefore,the study of a grading equipment for Nanfeng mandarin oranges,which is in line with the national conditions and can meet the actual production needs of farmers,and which can also predict the physical and chemical indexes,will be more conducive to improving the overall competitiveness of our fruit in the domestic and international markets.In view of the fact that the production of Nanfeng honey oranges in China is dominated by small-scale farmers,a machine vision grading device for the main quality indexes of Nanfeng honey oranges was developed,which is small in size,low in cost and equipped with two transport lines,and can simultaneously detect the size and color of Nanfeng honey oranges in two lanes and predict the soluble solids content.The main work and conclusions accomplished in this paper are as follows.1)Study the image processing methods of Nanfeng Tangerine.By comparing and analyzing various methods of image preprocessing and segmentation,it is found that the threshold segmentation effect is ideal and the processing speed is fast after the G channel and B channel difference of the image after Gaussian filtering.The minimum circumferential circle method was used to detect the size of Nanfeng tangerine,and the average absolute error was only 1.10 mm.2)Study on the prediction method of image physicochemical indexes of Nanfeng tangerine.Color histogram,color moment,color aggregation vector,Hu moment and gray level co-occurrence matrix were extracted to predict and model the soluble solids of Nanfeng tangerinium.The advantages of different modeling methods are compared and analyzed.Catboost regression effect is relatively stable,and the R2 of soluble solid model is 0.774.3)In view of the large volume and high price of the existing intelligent fruit grading equipment,the main quality grading device of Nanfeng tangerina is designed to meet the needs of small scattered farmers.In this design,two transportation lines are installed on the small light and simplified device to ensure efficiency.Designed nan fung orange grading system,choose the computer as the image processing system,information collection and image post-processing extracted from the rank and judged by the way of the queue record nan fung orange rating information and location information,the driver specified location electromagnet will nan fung oranges into the corresponding level in the stripper plate,and the overall classification rate can be up to 6 per second.The device has low noise,relatively stable operation,and little damage of Nanfeng tangerine.The classification accuracy of nanfeng tangerine fruit specification is 98%,and the classification accuracy of nanfeng tangerine soluble solid is 90%. |
