【摘要】 运动分割是计算机视觉领域的一个经典问题。高动态背景与相机运动是该任务的两个难点。空时显著性检测,已被应用到运动分割中,且被证明具有对高动态背景和相机运动鲁棒的特点。提出了一种基于混合动态纹理的空时显著性检测方法,并将其应用到运动分割之中。该方法首先利用混合动态纹理(MDT)方法对高动态背景进行建模,然后基于中心\邻域的框架计算显著性图。对显著性图进行适当的阈值处理,即得到运动分割结果。实验结果表明,提出的方法能够有效地改善运动分割的精度。更多还原

【Abstract】 Motion segmentation is a classical problem in computer vision research. Highly dynamic background and camera motion are two challenges ofthis task. Spatiotemporal saliency detection has been used in motion segmentation and is robust to the challenges. This paper presents a novelspatiotemporal saliency detection method based on mixtures of textures (MDT), which is then used to resolve the motion segmentation problem. Firstly,this paper models the highly dynamic background by mixtures of textures. Secondly, calculate the spatiotemporal saliency based on a center-surroundframework. Finally, the saliency map yields motion segmentation results after thresholding. The experiments demonstrate that the method can effectivelypromote the accuracy of motion segmentation.更多还原