基于全向声呐的引水隧洞点云模型去噪方法研究

Research on Denoising Method of Point Cloud Model of Water Tunnel Based on Omnidirectional Sonar

三维重建技术逐渐成为引水隧洞运营期安全检测的关键手段;而受隧洞特殊水文环境噪声、数据采集设备噪声以及载体运动噪声等影响,采集的点云数据不可避免的会遭受到噪声干扰,导致有用信息缺乏,不利于三维重建的进行;因此,文章提出了基于声呐数据特征点的点云去噪算法,实现隧洞点云数据的去噪;首先,该文依据引水隧洞声呐点云数据的特点,定义视觉距离和视角向量特征参数;其次,通过耦合视角向量与点云法向量估计点云漂移向量,并使用核函数方法估计视角距离参数的概率密度分布从而计算漂移距离;最后,采用漂移算法在保持点云模型特征的同时实现噪声的滤波;实验结果表明,文章提出的算法在去除隧洞点云模型数据噪声的同时能很好的保持引水隧洞模型的细节特征,为后续隧洞病害的检测提供高精度点云数据模型。

Three-dimensional reconstruction technology has gradually become a key means of safety detection during the operation of water diversion tunnel. Affected by the special hydrological environment noises of the tunnel, data acquisition equipment and carrier motion, the collected point cloud data will inevitably suffer from noise interference, resulting in a lack of useful information, which is not conducive to the three-dimensional reconstruction. Therefore, a point cloud denoising algorithm based on sonar data feature points is proposed to achieve the noise denoising of tunnel point cloud data. Firstly, according to the characteristics of the sonar point cloud data of the water diversion tunnel, the visual distance and perspective vector feature parameters are defined;Secondly, the point cloud drift vector is estimated by coupling the perspective vector and point cloud normal vector, and the probability density distribution of the perspective distance parameter is estimated by using the kernel function method to estimate the drift distance;Finally, the drift algorithm is used to achieve the noise filtering while maintaining the characteristics of the point cloud model. Experimental results show that the proposed algorithm can remove the data noise of the tunnel point cloud model while maintaining the detailed characteristics of the diversion tunnel model, and provide a high-precision point cloud data model for the detection of subsequent tunnel diseases.