图像视觉条状纹理矢量方向场检测技术

Vectorial Orientation Field Detecting of Striped Texture Image

方向场是条纹图像的重要特征量。本文提出了条纹图像矢量方向场模型,相比于通常的标量方向场,更符合方向的角度特性,并能通过矢量的模携带方向估计的锐度系数,使得对方向场的运算具有加权特性。同时,在矢量方向场的检测中,提出基元检测的概念。实验表明,通过建立矢量方向场,能够更准确地描述条纹图像的方向特征,并且有利于图像的后续处理。

Orientation is an important feature of the striped texture image. A conception of building a vectorial orientation field model of the striped texture image is presented in this paper. Compared to the traditional method, which defines orientation field as a scalar field, this method is more coincident with the angle character of orientation, also can take the acutance coefficient by the magnitude of vector, which could make the operation of orientation field with weight value. In addition, a conception of unit detection for orientation field estimation is introduced. The experiment indicates that the method of building vectorial orientation field can describe the directional character of striped texture image more exactly and is beneficial to processing image later.