基于边界矩和改进FCM聚类的水下目标识别

水下目标的识别是水下机器人对环境动态感知、快速定位与跟踪视觉目标的关键, 本文针对水下成像的特殊性以及成像环境的复杂性,旨在设计一种快速、准确的目标识别系统以指导水下机器人进行下一步的任务. 首先, 综合运用一些流行的算法, 简要介绍了一种有效的边界分割算法; 然后通过对边界矩的分析和修正, 构造了具有平移、旋转及比例变换不变性的仿射变换; 最后详细描述了改进的FCM聚类识别的设计理念. 通过对实测的4类物体组成的水下目标的识别实验, 证明了所提水下目标识别系统可以用于水下目标识别, 并且具有较高的鲁棒性和实时性.

Underwater targets recognition based on contour moment and modified FCM algorithm

The underwater targets recognition is the key to the dynamic perception of the environment, fast location and visual target tracking of autonomous underwater vehicle (AUV). To eliminate the negative effects, which are brought by the particularity and complexity of imaging environment, this paper presents a rapid and accurate target recognition system for AUV to use the obtained result for the next task. Firstly, an efficient contour segmentation algorithm is given by the integrated application of some popular algorithm. Secondly, affine transformation invariants which is constant to image translation, rotation and scaling are constructed through the analysis and modification of the contour moment. Lastly, the design idea of a modified FCM clustering recognition is discussed in detail. Experiments of underwater target recognition are made on four types of actual targets prove that the recognition system can be applied to underwater object detection and have the relatively high robustness and real-time effectiveness.