一种基于平稳小波变换的海底管线边缘检测方法

海底影像存在着对比度低、噪声污染严重、图像质量差等问题，采用传统算子的海底管线边缘图像中含有大量的无用和断裂边缘信息。文中将多尺度边缘检测和匹配跟踪相结合，提出利用平稳小波变换的海底管线边缘检测方法；在提取边缘的同时利用匹配跟踪手段对噪声干扰进行抑制，提高图像目标边缘检测质量。通过对海底管线和测试图像边缘检测实验表明，文中所提出的方法在抑制图像噪声的干扰、提高水下目标边缘完整性方面明显优于传统的边缘提取算子，证明该算法的有效性。

An Approach of Edge Detection of Subsea Pipelines with Stationary Wavelet Transform

It is well known that underwater images are inherently corrupted by multi noise sources so that the images are of low contrast and poor quality, The extracted edges of subsea pipeline images obtained with conventional edge detectors contain a great many useless and broken edges. In this paper, an approach is proposed by combining stationary wavelet transform and multi-scale edge tracker to denoise underwater images and to improve the quality of edge map for subsea pipeline image processing. The evaluation results on subsea images and testing images were demonstrated that the proposed approach has superior performances to conventional edge detectors in image denoising and edge detecion.