Abstract: In order to fast detect man-made objects in complex underwater scene, this paper presents a real-time straight line detection algorithm in the low frequency subband of wavelet transform of an image. The appropriate detection scale is first determined by using wavelet transform, and then edge detection is performed on the determined low frequency subband image to obtain the edge points corresponding to the significant line features, gradient histogram and iteration method are used to determine the adaptive edge detection segmentation threshold value. Secondly, the line features of man-made objects are detected by employing a refined Hough transform based method. Finally, the line detection results are marked on the original image. The experimental results show that the proposed algorithm can accurately find out the straight lines existed on man-made objects in complex underwater background and has favorable real-time performance, thereby satisfying the practical application requirement of underwater video based on man-made objects detection.

Key words: underwater man-made objects, edge detection, Hough transform, straight line detection, wavelet transform, iteration

摘要： 为了从水下复杂场景中快速检测人造目标，提出了一种在图像小波变换低频子带上进行直线实时检测的算法。首先，利用小波变换确定显著线特征检测的合适尺度；然后在确定的低频子带小图像上进行边缘检测，利用梯度直方图和迭代法相结合自适应确定边缘检测的分割阈值，得到显著特征的边缘点；再利用改进的Hough变换检测人造目标的直线特征；最后在原始图像上标记出直线检测的结果。实验结果表明：提出的算法可以准确检测出水下复杂背景中人造目标的直线特征，并且具有良好的实时性，满足水下人造目标视频检测的应用要求。

关键词: 水下人造目标, 边缘检测, Hough变换, 线特征检测, 小波变换, 迭代法