基于改进支持向量机的激光测距干扰数据分类

Jamming data classification of laser ranging based on improved support vector machine

传统采用机器学习方法对激光测距干扰数据实施分类,其将经验风险最小化当成优化目标,对海量点云数据的分类效果同真实结果间的偏差较高,受到非线性以及维数灾难的干扰,容易产生局部极小点问题。因此,提出基于改进支持向量机的激光测距干扰数据分类方法,基于激光点云对象原始特征,分析器邻近点群以及上下文语义环境,基于激光点云空间特征以及局部几何特征,获取点云对象的17个特征组成点云特征向量,进而塑造支持向量机分类模型,通过粒子群优化算法对该模型参数实施寻优操作,实现分类模型的改进,提高激光测距数据分类精度。采用改进支持向量机分类模型获取存在单类数据的最小球形边界,采用该边界对干扰数据实施分类,通过最小闭包球算法对支持向量机分类模型进行大数据分类优化求解,完成激光测距干扰数据的准确分类。实验结果说明,所提方法具有较高的适应度,分类效果好,取得令人满意的效果。

The traditional machine learning method is used to classify the laser ranging interference data,which sets minimizing the empirical risk as the optimization target. The classification effect of the massive point cloud data is higher than the real result,and is interfered by nonlinear and dimensional disasters and easy to produce local minimum points. Therefore,a laser ranging interference data classification method based on improved support vector machine is proposed. Based on the original features of the laser point cloud object,analyze the neighbor point group and the context semantic environment. The point cloud feature vector consisted of 17 features groups of cloud object is obtained based on the spatial characteristics of the laser point cloud and the local geometric features to shape the support vector machine classification model. The particle swarm optimization algorithm is used to optimize the model parameters to improve the classification model and improve the classification accuracy of laser ranging data. The improved SVM classification model is used to obtain the minimum spherical boundary of single-class data and to classify the interference data. The minimum closed-ball algorithm is used to optimize the big data classification of the support vector machine classification model to complete accurate classification of laser ranging interference data. The experimental results show that the proposed method has higher fitness,good classification effect and satisfactory results.