Target recognition and location based on binocular vision system of UUV
Target recognition and location based on binocular vision system of UUV
复制标题
DOI：
10.1109/chicc.2015.7260249
复制
发表时间：
2015-07
期刊：
影响因子：
--
通讯作者：
Xu Jian;C. Xiaoyuan;Song Xiaoping;Li Hang
中科院分区：
文献类型：
--
作者：
Xu Jian;C. Xiaoyuan;Song Xiaoping;Li Hang
针对无人潜航器在水下自主运行的特点，提出了一种改进的无人潜航器目标识别与定位算法。首先，利用Lenz畸变模型对双目摄像机进行标定；其次，在传统的高斯混合模型（GMM）中引入目标识别的可接受阈值，建立分类识别模型；通过将目标特征与模型进行比较，实现目标的识别。最后，利用NCC的分层匹配搜索方法对双目视觉图像中的特征点进行搜索和匹配，从而获得目标的位置和方向。坦克仿真结果表明：该方法可行、有效、实时性好，满足了无人潜航器自主作业的要求。
Since Unmanned Underwater Vehicle (UUV) operates underwater autonomously, an improved algorithm of object recognition and location for UUV is proposed in this paper. Firstly, the binocular camera is calibrated by Lenz distortion model; Secondly, the acceptable threshold for object recognition is used in the conventional Gaussian Mixture Model (GMM)to establish the classified recognition model. By comparing object features with the model, object is recognized. Finally, feature points in the binocular vision images are searched and matched using the hierarchical matching search method of NCC, thus the position and orientation of object is achieved. Tank simulation results have illustrated that: the proposed method is feasible, effective and real-time, which meets the requirements of autonomous operation by UUV.