基于三维彩色点云的物体识别算法

Object recognition algorithm based on 3D color point-cloud

随着新一代深度传感器的出现,使用三维(3-D)数据成为物体识别研究的热点,而且提出了很多点云特征描述子。针对传统的采用点云形状特征描述子在目标描述方面的不足,提出了一种基于三维彩色点云的物体识别算法。首先提取点云数据的视点特征直方图(VFH)和颜色直方图(CH),然后对提取的形状特征和颜色特征分别通过支持向量机(SVM)进行预分类,最后将上述2个识别结果进行决策级融合。提出的算法在Washington RGB-D数据集进行训练和测试。结果表明,该方法与传统的采用点云形状特征描述子相比,其物体的正确识别率有了显著的提高。

With the advent of new-generation depth sensors, the three-dimensional （3-D） data is used frequently on the object recognition, then a lot of point cloud feature descriptors are put forward. Based on the traditional using point cloud shape feature descriptors are insufficient, an object recognition algorithm based on the 3D color point cloud is proposed. First, the viewpoints histogram （VFH） and color histogram （CH） of a certain point cloud data is extracted. Then, the support vector machine is used to presort the extracteA features respectively. Finally, the above two recognition results is fused using the decision level fusion. The proposed algorithm is tested on the Washington RGB-D dataset. Experiment results show that the algorithm can effectively improve the correct rate of object recognition.