-
摘要: 利用局部邻域像素强度分布性质提出了一种快速局部特征描述算法---规范化强度对比描述子(Normalized intensity contrast descriptor, NICD). 首先规格化兴趣点邻域的像素强度, 再根据邻域中像素强弱分布建立描述子. 分别利用Fast-Hessian, DoG以及Harris-Laplacian检测子搭配NICD进行图像匹配以及物体识别实验. 结果表明: 在多种图像变换中, NICD可以实现与当前先进的SIFT和SURF算子相当的匹配效果, 而匹配时间大幅缩短, 因而更适合在实时应用中使用.Abstract: A fast local feature description algorithm is proposed in this paper based on the intensity distribution property of local neighbor, called normalized intensity contrast descriptor (NICD). After normalizing pixel intensity of local neighbor areas surrounding interest points, the descriptors could be computed based on the intensity distribution property of local neighbor. For evaluating the performance of NICD, we adopt three detectors: fast-Hessian, DoG, and Harris-Laplacian, with NICD respectively in image matching and object recognition test. The results of evaluations show that this new descriptor is competitive with the performance of SIFT descriptor and SURF descriptor in geometric and photometric deformations. However, the matching time is greatly shortened when using NICD. Therefore, NICD is more suitable in real-time applications.
-
Key words:
- Local feature /
- image matching /
- scale-space /
- object recognition
点击查看大图
计量
- 文章访问数: 2159
- HTML全文浏览量: 72
- PDF下载量: 1238
- 被引次数: 0