对称轴作为观测目标的一个重要特征,在计算机视觉中占有重要地位。提出的算法利用动力学中轴对称的刚体对于其对称轴的转动惯量取得极小值的特性,可以准确检测灰度图像中目标物体的对称轴。同时进一步提出了结合目标边界跟踪的对称轴检测方法。实验证明,本算法对于图像的尺度变化,旋转,平移都可以得到稳定的结果,具有良好的鲁棒性,实时性和准确性。
As an important character of target, symmetry axis have a high status in computer vision. In this paper,a symmetry axis detected arithmetic is introduced based on the theory in dynamics that a rigid body have the minimal value of moment of inertia to its symmetry axis. Furthermore, this paper presented a new method combined with outside boundary tracking. The experiment' s result proved that method could detect the symmetry axis of the target in the gray - scale image robustly, exactly and fastly.