Abstract: An advanced chord midpoint Hough transform ellipse detection method is proposed for geometric characteristics of underwater objective to extract feature of multi-objective. Using beam interpolating method and coordinate transform method for image data in two-dimension polar form received by sonar to generate image, it makes pre-processing and contour tracking, and extracts feature of multi-objective. The experimental results show that the method works well for multi-objective feature extraction, and provides useful information for multi-objective recognition and positioning.

Key words: forward-looking sonar, Hough transform, ellipse detection, multi-objective feature extraction, image pre-processing, contour tracking

摘要： 针对水下目标图像的几何特性，提出了一种基于弦中点的改进Hough变换椭圆检测方法，实现了水下多目标的特征提取。单波束机械扫描前视声纳的图像数据为二维极坐标形式，采用波束内插方法和坐标变换方法生成图像，并对其进行图像预处理和轮廓跟踪，据此再进行多目标特征提取。通过水池实验结果表明，该方法对于水下多目标的特征描述更简明准确，为多目标识别与定位提供了有用信息。

关键词: 前视声纳, Hough变换, 椭圆检测, 多目标特征提取, 图像预处理, 轮廓跟踪