基于水平集算法后向散射的运动目标轮廓提取

基于距离选通的激光主动成像系统在大气条件下对运动目标成像时，为了在更大范围内搜索跟踪目标，距离选通初始门宽通常在微秒量级，进而引起后向散射，严重影响了系统对目标的识别与提取。文中首先分析了基于距离选通的激光主动成像系统，在此基础上引入距离信息实时跟踪运动目标，研究了不同选通门宽引起的后向散射对图像质量的影响，提出了一种基于Sobel 算子和 GAC 模型的目标轮廓提取算法，将 Sobel 边缘检测的结果引入水平集边界停止函数，为水平集的曲线演化提供参考依据。实验结果表明，提出的水平集算法能够克服严重后向散射干扰影响提取目标轮廓，目标提取精度在0.9以上，高于 GAC、LIF、LBF 等经典水平集算法。

Level set algorithm for the extraction of target contour under backscatter interference

When the range-gated laser active imaging system is imaging the moving targets under atmospheric conditions,the initial gate width is usually in the microsecond magnitude in order to search and track the target in a wider range,which causes much backscatter and seriously affecting the recognition and extraction of the target.The range-gated laser active imaging system was analyzed firstly,then real-time tracking of moving targets from the motion information was introduced.The effect of the backscatter caused by different gate widths on image quality was studied,and a kind of object contour extraction algorithm based on Sobel operator and GAC model was put forward.The result of Sobel edge detection is introduced to level set boundary stop function,which provides a reference for the curve evolution of level set.Experimental results show that the level set algorithm can overcome the effect of serious backscatter on object contour extraction,and target extraction accuracy is above 0.9 and higher than the GAC,LIF and LBF classic level set algorithm.