一种面向图像线特征提取的改进投票域的张量投票算法

A tensor voting algorithm for image line feature extraction based on improved voting field

张量投票算法利用人类感知功能原理进行计算,它具有较强的鲁棒性、非迭代性、参数唯一性等特性,其非迭代性具有节省计算时间的显著性特征,因此,广泛应用于图像线特征提取,但在一些含有复杂噪声的图像中,却不能得到更为连续的显著线特征信息。本文针对此问题,提出一种改进的具有迭代性的张量投票算法,它主要是对投票域进行迭代改进,使改进后的张量投票算法可以提取更为连续的显著线特征,且与传统的张量投票算法相比,本文算法既缩短了计算时间,又提取了更为连续的线特征图像。

Tensor voting algorithm is calculated by the principle of human perception function,which has strong robustness,non-iteration,uniqueness of parameters and other characteristics.Its non-iteration has the significance feature of saving computation time,so,it is widely used in image line feature extraction.However,in some complex noisy images,it cannot obtain more continuous saliency line feature information.In order to solve this problem,an improved iterative tensor voting algorithm was proposed,which mainly improved the voting domains iteratively,so that the improved tensor voting algorithm could extract more continuous salient line features.Compared with the traditional tensor voting algorithm,the proposed method in this paper not only shortened the calculation time,but also extracted more continuous line feature images.