一种基于相似运动模式的群体分割方法

由于内部复杂性、多样性等因素存在，复杂场景下基于相似运动模式的群体分割一直是计算机视觉领域中的难点问题。为了提高群体分割结果的准确性，本文将高精度变分光流模型与基于拉格朗日流体动力学的脉线模型相结合，提出了一种高精度的群体分割算法。该算法首先采用各向异性扩散模型对原始视频帧进行平滑，保证了去噪图像的重要结构信息；接着将高精度的变分光流模型与脉线模型结合，得到抗干扰能力较强、精度较高的脉线（streakline）和脉线流；然后根据脉线和脉线流的相似性采用分水岭算法对群体进行分割；最后通过对多组视频序列进行试验对比分析，验证了该方法的有效性和准确性。

A Crowd Segmentation Method Based on Similar Motion Pattern

Crowd segmentation based on similar motion patterns in crowded scenes remains an open problem in computer vision due to inherent complexity and vast diversity found in such scenes. To improve the accuracy of the crowd segmentation, a method combining the high accurate variational model with the streakline framework based on Lagrangian Fluid Dynamics is proposed. First, in order to keep the important structure of information of denoising image, the anisotropic diffusion model is used to smooth each frame of the video. Second, the streaklines and streak flow with high accuracy and strong anti-interference ability are obtained by combining the variational model with the streakline framework. Third, each frame of the video is segmented into different regions based on the similarity of streaklines and streak flow using watershed segmentation. Last, experiments are done to verify the validity and accuracy of our method.