一种基于支持向量机的孔洞修补方案评估方法
|
摘要
为了自动评估逆向设计软件的点云孔洞修补效果,以确定最佳的修补方案,提出了一种基于支持向量机的点云孔洞修补评估算法。首先,以基于实际工程应用的挖掘机斗齿点云为实验对象,在斗齿的不同部位人为构造20个点云孔洞,用专业逆向设计软件Imageware13.2进行孔洞修补,并建立相关特征向量。其次,将60%的特征向量作为训练样本输入SVM模型进行训练,并建立参数优化后的SVM分类器模型。最后,将其余特征向量作为测试样本输入SVM模型进行分类识别。实验证明,此方法具有较好的识别效果,识别正确率可达90%,并有鲁棒性好、识别速度快等优点。
This paper presents a novel hole-recognition algorithm in reverse engineering(RE) that can automatically evaluate the effectiveness of the software based hole filling method thus to determine the best filling schemes for holes in the point clouds. First, 20 manmade holes were circle selected from different portion of a real-world scanned objects-a bucket of an excavator, and then all of them were reconstructed with Imageware 13.2. Second, 60% of them were used to train the SVM and to establish the pattern recognition model, while others were for SVM predicting. The experimental results show that the suggested approach performs quite well, it is efficient, robust and what's more, the average recognition accuracy can be up to 87.5%.
This paper presents a novel hole-recognition algorithm in reverse engineering(RE) that can automatically evaluate the effectiveness of the software based hole filling method thus to determine the best filling schemes for holes in the point clouds. First, 20 manmade holes were circle selected from different portion of a real-world scanned objects-a bucket of an excavator, and then all of them were reconstructed with Imageware 13.2. Second, 60% of them were used to train the SVM and to establish the pattern recognition model, while others were for SVM predicting. The experimental results show that the suggested approach performs quite well, it is efficient, robust and what's more, the average recognition accuracy can be up to 87.5%.
引文
[1]Yann Quinsat,Claire lartigue.Filling Holes in Digitized Point Cloud Using a Morphing-based Approach to Preserve Volume Characteristics[J].Int J Adv Manuf Technol,2015,(81):411-421.
[2]晏海平,吴禄慎.基于径向基函数的散乱点云孔洞修复算法[J].计算机工程与设计,2014,35(4).
[3]刘咏梅,李凤霞.基于边界特征增长的孔洞修补算法[J].系统仿真学报,2014,26(9).
[4]Fortes M A,Gonz′alez P.Filling Holes with Shape Preserving Conditions[J].Mathematics and Computers in Simulation,2015,(118):198-212.
[5]王春香,陈浩宏,孔艳彬,等.RBF神经网络在斗齿散乱点云漏洞修补中的应用[J].机械科学与技术,2014,33(10).
[6]岳雷,沈庭芝.基于自动提取特征点的三维人脸表情识别[J].北京理工大学学报,2016,36(5).
[7]Caputo1 M,Denker1 K.Learning Geometric Primitives in Point Clouds[J].Euro graphics Symposium on Geometry Processing,2014.
[8]常伟,杰蔡勇.基于SVM的点云模型识别技术的研究[J].机械设计与制造,2009,9(9).
[9]曹林,朱国刚.基于三维时空直方图特征的人体行为识别[J].计算机工程与设计,2016,37(4).
[2]晏海平,吴禄慎.基于径向基函数的散乱点云孔洞修复算法[J].计算机工程与设计,2014,35(4).
[3]刘咏梅,李凤霞.基于边界特征增长的孔洞修补算法[J].系统仿真学报,2014,26(9).
[4]Fortes M A,Gonz′alez P.Filling Holes with Shape Preserving Conditions[J].Mathematics and Computers in Simulation,2015,(118):198-212.
[5]王春香,陈浩宏,孔艳彬,等.RBF神经网络在斗齿散乱点云漏洞修补中的应用[J].机械科学与技术,2014,33(10).
[6]岳雷,沈庭芝.基于自动提取特征点的三维人脸表情识别[J].北京理工大学学报,2016,36(5).
[7]Caputo1 M,Denker1 K.Learning Geometric Primitives in Point Clouds[J].Euro graphics Symposium on Geometry Processing,2014.
[8]常伟,杰蔡勇.基于SVM的点云模型识别技术的研究[J].机械设计与制造,2009,9(9).
[9]曹林,朱国刚.基于三维时空直方图特征的人体行为识别[J].计算机工程与设计,2016,37(4).