高级检索

    基于块运动矢量加权的Snakes模型及在序列图像分割中的应用

    Sequential Image Segmentation Using Block-Motion Vector Weighted Snakes Model

    • 摘要: 经典的Snakes模型具有开放的、统一的架构 ,在此基础上 ,为了分割复杂背景的序列图像 ,产生了各种改进的Snakes模型 ,但都存在着不足 :计算量大、需要先验知识、易受光流计算精度影响等 针对这些缺点 ,提出了块运动矢量加权的Snakes模型 ,可以用于复杂背景序列图像的分割 这种模型以图像中的边缘信息为分割的最终依据 ,结合块运动估计的结果 ,增强了序列图像分割的鲁棒性 根据运动场估计的结果在该模型中所起的作用 ,提出了边缘优先的块运动估计算法 ,大大减少了计算量 用块运动矢量加权的Snakes模型分割复杂背景序列图像 ,取得了好的分割结果

       

      Abstract: Snakes model ,with an open and uniform framework, is an elegant tool for image segmentation The active contour can converge dynamically to the boundary of the moving object in a sequential image by constructing an external force properly At present, many modified snakes models are presented for sequential image segmentation, but they have some deficiencies such as high computational complexity, requiring prior knowledge, being limited by the precision of optical flow field and so on The Snakes model presented overcomes these shortcomings, and can segment sequential images effectively The information of edge is vitally important for image segmentation, so it is the foundation of constructing external force In the mean time, the kinetic information, obtained by motion estimation, is incorporated into sequential image segmentation, and therefore, the novel Snakes model is applicable for the sequential image with complex background Furthermore, an edge first block based motion estimation is proposed according to its role in the block motion vector weighted Snakes model, which reduces the computational load evidently Finally, the promising experimental results are provided to illustrate the validity of the block motion vector weighted snakes model for sequential image segmentation

       

    /

    返回文章
    返回