Research On Log-end Area Recognition Methods Based On Cluster Analysis

The forest resource is an indispensable part of the development of ecological civilization and the survival of human society.At present,people's quality of life requirements rise,and the demand for forest resources mainly from two aspects.First,all walks of life on the demand for logs and products increased year by year,and second,people are increasingly demanding the ecological environment,want to have more and better forest resources.So,in order to ensure the improvement of economic and ecological benefit,we must make sure the limited forest resources are scientifically utilized and effectively protected.Growing on mountain,timber only produce economic benefit after transported to users.Then,timber transportation management is an important part of forest resource protection and management.But Manual gauging precision is not high and low efficiency,and easily affected by the subjective consciousness of employees.Therefore,it is very important to realize the automatic log gauging.The automatic log gauging can improve the efficiency of the inspection work and the illegal transport problems in the process of timber transportation,and can play an important role in forest resource protection.In the view of the above problem,this paper mainly studies on log-end area recognition based on clustering.The purpose is to extract the log-end area from the image,laying a foundation for the realization of the automatic log gauging.First,this paper introduces the current status of the log-end area recognition in the country and the research status of image processing based on cluster analysis.Secondly,the theoretical basis of K-means algorithm,FCM clustering and CFSFDP algorithm is introduced,and the details of the three clustering algorithms and the concrete operation steps are described in detail.At the same time,it also introduces the theoretical basis of color space and multi-scale multivariate image analysis in the image processing.Thirdly,the K-means algorithm,FCM clustering and CFSFDP algorithm are applied to the log area recognition of logs,and the experimental results are analyzed.Experiments show that the CFSFDP algorithm with multi-scale multivariate image color characteristics has the best experimental results in these three clustering algorithms.CFSFDP algorithm not only the image with simple background can achieve good results;the image with more complicated background also can get ideal result.Finally,for the log contour recognition,the Hough transform can solve the problem that the log contour deviates from the intersection of the circle and the contour,in the view of the circle nature of the logs,and the contours between the logs and the logs are interconnected.Therefore,on the basis of extracting the log-end area,the log contour recognition can be realized by Hough transform circle detection.According to the investigation of the main forest area in Fujian Province,it is urgent to design the automatic number recognition system of logs,which requires 90%recognition accuracy.Therefore,this paper proposes the method of using the CFSFDP algorithm and the multi-scale multivariate image analysis method to obtain the log end area of the log,and then use the Hough transform to realize the recognition of the contours of the logs in the image.The method is designed as a software system and can be used for the actual timber transport management process.