针对水下遥操作作业时目标深度定位及水下复杂环境成像困难的问题, 利用基于相机阵列的信息获取系统, 使用基于深度的集成成像三维重建算法及相邻像素灰度方差评价函数, 实现了一次处理可同时实现有遮挡情况下目标的深度估计及清晰成像。应用LabVIEW软件编程建立了CCD相机自动扫描阵列成像系统, 对存在遮挡物的水中目标场景进行成像及深度重建实验, 场景中位于不同深度的3个独立目标重建深度值与实际深度值的误差分别为0.21%、1.26%和1.34%。实验结果表明, 应用该方法仅需一次重建即可去除遮挡物, 并获得目标的深度值。

To solve the problems of depth location of underwater targets and complex underwater imaging environment during underwater teleoperation, an information acquisition system based on the camera array and an integral imaging three-dimensional reconstruction algorithm based on depth and sum modulus difference evaluation function are used to achieve the target depth estimation and clear imaging with occlusion simultaneously. A CCD camera automatic scanning array imaging system is established by the programming of LabVIEW software. Imaging and depth reconstruction experiments of occluded underwater targets are carried out. Errors of three independent targets′ reconstruction depths and the actual depths at different depth are 0.21%, 1.26% and 1.34% respectively. Experimental results show that this method can remove the occlusion and obtain the depth value of target simultaneously after only one reconstruction.