几何活动轮廓模型在目标跟踪中的应用

针对复杂的视频场景中目标追踪易受环境干扰的问题，提出了一种基于混合高斯模型和改进的C-V（Chan-Vese）模型相结合的新方法。其中采用了混合高斯模型算法更新背景，检测出运动目标轮廓。然后对提取出的目标轮廓进行后处理，标定出运动目标的质心和运动区域。将运动区域作为初始化曲线，用改进的C-V模型对运动目标进行拟合。结果证明了以标定出的运动目标区域为初始化曲线可以有效地提高轮廓曲线的收敛速度；对于灰度不均匀的和含有噪声的图像，改进的模型的分割效果也要好于C-V模型和LCV模型。

Application of geometric active contours in target tracking

Aiming at the problems that the target tracking is easily disturbed by the environment, this paper proposes a new algorithm based on Gaussian mixture model and the improved C-V（Chan-Vese） model. Firstly, the paper uses Gaussian mixture model to update the background and detect the contour of the moving target. Then, it does post-processing on the extracted target contour and calibrates the center of the moving target and the moving target region. Finally, the moving target region is used as the initialized curve and the improved C-V model is proposed to fit the moving target. Experimental results demonstrate that using the calibration movement target area as the initialized curve can effectively improve the convergence rate of the contour curve. Besides the segmentation of improved model is better than C-V model and LCV model in the uneven gray image and noise image.