形状约束的Snake算法在探地雷达图像目标自动提取中的应用

对探地雷达图像进行处理，提取目标的精确轮廓，从而正确估计出地下管线的管径、埋深等参数是自动定位的关键技术。考虑到地下管线的回波形状是已知的，理论上为双曲线，作者采用加入形状约束的改进Snake算法（Active Contour Model)对探地雷达图像进行目标提取，并对其在实用中不稳定的特点，做了进一步改进。对于Snake算法的初始包络自动提取，作者在文中也提出了一种新的方法。实验结果表明本方法迅速可靠，已应用于实际的探地雷达中。

THE AUTOMATIC TARGET DETECTION ON GROUND PENETRATING RADAR IMAGES USING SHAPE-BOUNDED ACTIVE CONTOUR MODEL

This paper presents a new method for automatic detection of the pipeline boundary on ground penetrating radar images using an active contour model. The method of extracting objective boundary accurately is a key technique for the automatic determination of a pipeline location. Considering that the echo wave shape of an underground pipeline is hyperbolic in theory, this paper studies a shape bounded active contour model for the detection of a pipeline boundary on the ground penetrating radar image and presents a new algorithm for the automatic detection of the pipeline boundary, in which the initial contour is extracted without user's interaction. The instability of the active contour model in application is also improved. The authors also suggest a new algorithm for the automatic determination of the initial envelope of the active contour model. The experiment results show that the algorithm is fast and reliable, and now it has been applied to an actual ground penetrating radar image processing case.