Research On Camera Calibration Method Based On Standing Tree Measurement

Pruning is one of the most important and necessary step of cultivating plantation forest.Machine vision for automatically pruning can enhance the pruning machine efficiency,achieve intelligent and automation operation.Diameter measurement of the standing tree is an important part of the machine vision for automatically pruning.According to the current trunk diameter measurement tools,based on the camera calibration method,using image process and neural network technology,this paper gives a automatic standing tree trunk diameter measurement method based on the neural network camera calibration.The aim of this method is image process and feature detection for the trunk,analysis of the specific camera model and calibration method in order to improve calibration of BP neural algorithm,achieving the automatic tree trunk diameter measurement.In this paper,the results achieved and the innovation as follows:1,In the calibration experiment,take the special positions to take the pictures of calibration board after referring experiment design of the camera calibration based on the 2D plane board,after the special points computing,the external parameters[R,t]is got, this can simplify the ordinary camera calibration computing and give the reliable data to the neural network calibration;2,Through the research of traditional method of camera calibration,the distance factor between the camera and the target is proposed,BP neural network algorithm is used and finished the single camera calibration model.Compare with the common method, this method can overcome the camera distortion and more easy to get.3,According to standing tree trunk image features,the paper find a suitable detection method for standing tree trunk edge.First,using the threshold segmentation to process the gray image,then using the median filtering and the elimination the isolated points to remove the majority part of noise,and then using the expansion and corrosion to integrate the white points,finally using the Krisch method to get the closed curve of the trunk edge.This method combines the characteristic of the standing tree trunk,the steps is relatively simple,ensuring the original state of the tree.4,Based on the BP neural network calibration of the single camera,using BP neural network to measure the diameter of the tree trunk is proposed,this method can realize the automatic measuring the diameter of the tree trunk,the measurement distance is between 0.5m-1.25m.The experimental results show that this method can achieve the measurement accuracy 0.1mm,measurement error is +/-2.80 mm,this can meet the measurement requirements.