基于图像融合的运动目标轮廓提取新方法

A Novel Contour Extraction Algorithm for a Moving Target Based on Image Fusion

提出了一种新的运动目标轮廓提取方法。该方法分为3个步骤:对于红外与可见光图像序列,首先应用运动分割得到运动目标的初始轮廓线;其次,使用了动态轮廓线使其收敛到目标轮廓;最后,应用了两种新的图像融合方法,第1种方法求解两类模式图像的动态轮廓线控制点向量差的L2范数平方极小化,第2种方法通过图像配准实行改进的微分耦合。对比检测实验表明,第2种方法具有较好的融合效果。同时,设计了一种基于Newmark方法的动态轮廓线快速迭代算法,该方法如Wilson-θ方法的对比实验表明,Newmark方法稳定收敛的迭代时间降低了21.01%。

A novel contour extraction algorithm for a moving target is proposed. This algorithm consists of three steps. Firstly, motion segmentation is applied respectively to infrared and visible image sequences to get an initial contour of a moving target. Secondly, a dynamic contour is applied to make the initial contour converge to the target＇s contour. Finally,two novel image fusions are applied. The first fusion minimizes L2 norm＇s square of the difference of control point vectors in two modal images without image registration. The second fusion is realized by the modified differential coupling with image registration. Contrasting detection experiments indicate the second fusion is better. Meanwhile,a fast iteration algorithm of dynamic contour based on the Newmark method is devised. A contrasting experiment indicates its iteration time for stable convergence decreases by 21.01% compared with the Wilson-θ method.