基于双目视觉的水下动态目标识别与定位方法研究

针对动态水下目标跟踪定位过程需要良好的实时性与鲁棒性问题，提出一种基于双目视觉的水下动态目标定位方法；利用快速引导滤波提取水下图像的光照分量，构造了一种改进的二维伽马函数，并对其参数利用光照分量的分布特性进行调整，实现了对水下不同环境图像下的亮度自适应校正处理；利用卡尔曼滤波预估下一时刻目标的位置，将预测空间作为ROI区域进行图像校正，极大降低了算法的运行时间；在HSV空间对目标进行掩膜提取，识别之后通过双目定位算法对目标进行准确定位；经过水箱试验验证，与多尺度高斯函数、双边滤波等算法相比，该方法在运行速度上有着显著的提高，达到了35FPS,在定位过程中有着较高定位精度，在方向的平均相对误差为(3.59%,3.35%,1.42%);结果表明，该算法可以满足水下动态目标跟踪定位的实时性与鲁棒性要求。

Research on underwater dynamic target recognition and location method based on binocular vision

Aiming at the problem that the dynamic underwater target tracking and positioning process requires good real-time and robustness, a method of underwater dynamic target localization based on binocular vision is proposed. Firstly, the illumination component of the underwater image is extracted by fast guidance filtering, and then the two-dimensional gamma function is constructed and the parameters of the two-dimensional gamma function are adjusted by using the distribution characteristics of the illumination component, so as to realize the adaptive correction processing of the uneven illumination image. Secondly, the Kalman filter is used to predict the position of the target at the next moment, and the prediction space is used as the ROI region for image correction, which greatly reduces the running time of the algorithm, and then the target is masked in the HSV space, and the target is accurately located by the binocular positioning algorithm after recognition. Finally, compared with multi-scale Gaussian function and bilateral filtering algorithms, the proposed method has a significant improvement in running speed and high positioning accuracy in the positioning process, and the results show that the algorithm can meet the real-time and robustness requirements of underwater dynamic target tracking and positioning.