Research On Weed Classification Based On K-MEANS Algorithm

With the progress of the times,k-means algorithm has made breakthroughs in the integration of technology in various fields,especially in image recognition,playing a key role.Through the research and analysis of K-means algorithm,this paper applies it to weed image processing,clustering and putting the processed weed image data into a database,and comparing with the newly inputted images to determine the classification and attribution of weeds.By improving the shortcomings of K-means algorithm,a hierarchical clustering is added before the whole operation,which optimizes the problem of time-consuming and low accuracy of the classical algorithm in the whole weed classification process.It provides a simple and fast method for weed classification and weed removal.Specific research contents are as follows:1.By studying the current weed classification technology and the properties of K-means algorithm,this paper explores the methods and ideas of weed classification application in this paper.This paper studies the application of K-means algorithm in the field of face recognition,understands various weed classification techniques,and analyses the history,nature,definition,key points and application of K-means algorithm.2.The k-means algorithm is optimized.On the basis of fully understanding the k-means algorithm and analyzing the shortcomings of the K-means algorithm,a hierarchical clustering method is proposed before the k-means algorithm is implemented,which improves the accuracy of the classical algorithm and reduces the operation time.3.Through the experiment of the improved algorithm,weeds are classified quickly and simply.In order to verify the improved algorithm and the correctness of the research ideas of weed classification,a complete experiment was carried out.Four kinds of weed pictures were extracted and clustered into a database.New pictures were input to repeat the above process and weed categories were determined.