Affine Invariant Matching Method for Image Contains Repetitive Patterns
- Resource Type
- Conference
- Authors
- Wang, Yunshu; Liu, Jianye; Zeng, Qinghua
- Source
- 2016 International Conference on Digital Image Computing: Techniques and Applications (DICTA) Digital Image Computing: Techniques and Applications (DICTA), 2016 International Conference on. :1-6 Nov, 2016
- Subject
- Communication, Networking and Broadcast Technologies
Computing and Processing
General Topics for Engineers
Geoscience
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Pattern matching
Context
Matrix decomposition
Testing
Robustness
Lighting
Shape
- Language
Image contains repetitive patterns always cause the point ambiguity, which makes the local descriptors less discriminative. The descriptor which is combined scale-invariant feature transform (SIFT) with the global context (GC) is used to solve the problem widely. But this descriptor is invalid when the variation of viewpoint is large. In this paper, an affine invariant matching method for image contains repetitive patterns is proposed. The similar maximally stable extremal regions (MSERs) based on the repetitive patterns are detected. The affine matrix of the similar MSER is calculated and the image is transformed by the affine matrix to decrease the change of the viewpoint. Then the keypoints in the transformed image are descripted by the descriptor which is combined the SIFT with the improved GC. The GC is improved by using difference curvature instead of principle curvature value. The experiments show that the proposed method is valid for the images with large variation of viewpoint while the SIFT+GC is invalid. The proposed method also has the higher precision than ASIFT+GC while the time cost is less than half of ASIFT+GC.