网络数字图像边缘高精度配准方法仿真

Research of Network Digital Image Edge and High Precision Matching Method

为提升对网络数字图像的识别质量,需要进行图像边缘高精度配准方法的研究。但是采用当前方法进行图像边缘配准时,难以提取精确的图像边缘特征,存在配准误差大的问题。为此提出一种基于改进ORB的网络数字图像边缘高精度配准方法。上述方法先利用SUSAN算法提取图像边缘线,利用几何重心将图像边缘分为封闭边缘线和非封闭边缘线,计算图像边缘的大小极值点,提取图像边缘特征,并利用OLB描述子对图像的边缘特征进行描述,通过Hamming距离完成对数字图像边缘高精度的配准。仿真证明,所提方法配准精度高,为网络数字图像的识别提供了重要的价值。

A high precision matching method for network digital image edge based on modified ORB is proposed. Firstly, the SUSAN algorithm is used to extract edge line of image, then, the geometric center of gravity is used to di- vide image edge into close edge line and unclosed edge line, and the extreme point of image edge size is calculated. Moreover, the feature of the image edge is extracted, and the OLB descriptor is used to describe edge feature of image. Finally, the high precision matching is completed via Hamming distance. Simulation proves that the method has high matching precision. It can provide important value for recognition of network digital image.