Research On Fruit Type Recognition Based On Computer Vision

In recent years,with the development of computer technology,machine vision has been widely used in various fields.Among them,the use of machine vision technology to accurately identify fruit types has been a research hotspot.Its purpose is to realize automatic fruit type identification,save agricultural labor resources,improve production efficiency,and promote the development of smart agriculture.At present,many scholars at home and abroad have used pattern recognition methods such as support vector machines(SVM)to identify fruit types by extracting multiple characteristics of fruits.These recognition methods all use only one classifier,and the recognition effects of different classifiers on different fruit types are not balanced.Aiming at the problem of uneven recognition of different fruit types by different classifiers,this paper proposes a fruit recognition method based on the fusion of multi-classifier DS evidence theory.It mainly includes four contents: collection of a variety of fruit images,preprocessing of the collected images,feature extraction of preprocessed images,and design of recognition algorithms.The thesis first collects the research objects,and selects five fruit images on the fruits360 data set on kaggle as the research objects.Next,preprocess the collected images to remove the noise in the image collection process,and minimize the impact of data deviation caused by noise on the subsequent feature extraction.Then,extract the three features of color,texture,and shape from the preprocessed image.Finally,a fruit recognition algorithm combining the DS evidence theory and BP neural network,K-means,and SVM three classifiers is designed.The test combines the recognition results of the tested images on each classifier and the classification of different fruits by each classifier.Classification accuracy,construct the basic probability function of DS evidence theory(BPA function),and obtain the recognition result of the tested image after fusing the three classifiers through DS evidence fusion rules.The test results show that the average accuracy of the method for the recognition of five kinds of fruits is 95.2%,and the overall standard deviation is 0.02993.While improving the accuracy of single classifier recognition,it also solves the problem of uneven recognition of various fruits by the classifier.The average accuracy of the recognition of the 10 test sets is 93.5%,and the overall standard deviation is 0.055.This method combines the effects of different classifiers on the recognition of different fruits,and achieves a relatively stable and accurate recognition effect.