一种基于图像融合的运动目标快速检测算法

提出了一种新的运动目标检测方法,这种方法可以有效的提取目标轮廓。应用一种图像差分技术得到运动目标的初始轮廓线。使用了动态轮廓线使其收敛到目标轮廓。提出了一种新的目标轮廓特征级融合方法,求解两类模式图像的收敛动态轮廓线控制点向量差的范数平方极小化。这种方法不需要图像配准降低了融合的计算复杂度,有效提高了可见光图像中目标轮廓提取的精度。对比检测实验证实了算法的有效性。设计了一种基于Newmark方法的动态轮廓线快速迭代算法,将该方法和方法作了比较,对比实验表明这种方法的时间复杂度降低了22%。

Fast detection scheme for a moving target based on image fusion

A new detection scheme was proposed, which can extract a moving target's contour effectively. An image difference technology was applied to get the initial contour. Dynamic contour was used to make it converge to the target's contour. A new feature level fusion method of target's contour was proposed, which minimizes norm's square of the difference of control point vectors of convergent dynamic contours in two modal images. Image registration is not needed so that computational complexity is decreased for the fusion. This fusion improved the accuracy of contour extraction in a visible image effectively. Contrasting detection experiments proved the efficacy of the fusion scheme. A fast iteration algorithm of dynamic contour based on Newmark method was devised, which was contrasted with Wilson-θ method. A contrasting experiment indicated that its temporal complexity decreases by 22%.