This paper presents an efficient scheme for recognizing the fork branches by using support vector machines in order to meet the need of the machine vision system of the intelligent pruning robot. In this study, the diameters of trees and branches are measured by using the method based on surveyor's object. And then whether the branches are fork branches is estimated by support vector machines, according to the diameters of trees and branches. In this study, 489 two-dimensional data are obtained in all, which are composed of the diameters of trees and branches. 12 two-dimensional data that are selected at random are used to cross validate and train the support vector machine. The correct rate of support vector machine that is tested by 14 two-dimensional data is 100%, the percentages of sensitivity (SE), and specificity (SP) are all 100%. To the 463 two-dimensional data, The CC percentage is 96.11%, while SE is 100%, and SP was obtained as 95.93%. Experimental results demonstrate that the scheme is effective in recognition of the standing tree branches, and a foundation of the in-depth research on the machine vision system of the pruning machine.
[1]
Zhang Junmei.
Development on Wireless Remote Control System of Pruning Robot of Commercial Plantation
,
2003
.
[2]
Bernhard Schölkopf,et al.
Comparing support vector machines with Gaussian kernels to radial basis function classifiers
,
1997,
IEEE Trans. Signal Process..
[3]
Bernhard Schölkopf,et al.
Support Vector methods in learning and feature extraction
,
1998
.
[4]
Zhang Xuegong,et al.
INTRODUCTION TO STATISTICAL LEARNING THEORY AND SUPPORT VECTOR MACHINES
,
2000
.
[5]
Bernhard Schölkopf,et al.
Nonlinear Component Analysis as a Kernel Eigenvalue Problem
,
1998,
Neural Computation.
[6]
Russell M. Mersereau,et al.
Blur identification by the method of generalized cross-validation
,
1992,
IEEE Trans. Image Process..