| 摘要: |
| 针对ORB特征描述算法没有解决尺度不变性的问题，提出了一种将具有尺度不变性的BRISK特征描述符与ORB特征检测子相结合的特征点匹配算法；利用ORB特征检测子检测待测图像中的特征点，并借鉴BRISK特征描述算法的思想对检测到的特征点进行均匀采样，然后对采样到的特征点进行特征描述，最后使用暴力匹配的方法计算汉明距离从而完成特征匹配；通过实验验证了改进算法，有效地解决了ORB特征描述算法不具备尺度不变性的问题，相较于原ORB算法，改进算法的尺度不变性得到了有效提高且更加稳定、可靠，同时，改进算法的实时性也略优于原算法，适合于要求实时性高且尺度变化大的应用中。 |
| 关键词: ORB BRISK 尺度不变性 特征点匹配 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| An Improved ORB Feature Points Matching Algorithm |
|
ZHANG Yang, XU Gang, ZHANG Xing yu, JIANG Juan juan
|
| Abstract: |
| Aiming at the problem that the ORB feature description algorithm does not solve the scale invariance, a feature point matching algorithm combing BRISK feature descriptor with scale invariance and ORB feature detector is proposed. Firstly, the feature points in the image to be measured are detected by ORB feature detectors with scale invariance. Secondly, uniform sampling for detected feature points is conducted by referencing BRISK feature points description algorithm. Finally, the method of BF matching is used to calculate the Hamming distance for feature matching. The experimental results show that the improved algorithm effectively solves the problem that ORB feature description algorithm does not have scale invariance. By comparing with original ORB algorithm, the scale invariance of the improved algorithm is effectively improved and becomes very stable and reliable, simultaneously, the real time performance of the improved algorithm is better than the original algorithm. It is suitable for the applications that require high real time and large scale changes. |
| Key words: ORB BRISK scale invariance feature point matching |
