원문정보
초록
영어
In this paper, the natural features extracted from images do not use the specific markers or reference points proposed augmented reality system. First, for tracking objects in video footage after the creation of feature extraction and descriptors in the same way about the creation of feature extraction and descriptors will. Descriptor information and features I1I2I3 color around the edge of space, size and rotation is created by using the information. This generated using the descriptors for each feature determine the similarity and differences between the values obtained using the distance to check for matching. This process is generated through the corresponding points for two-dimensional plane to obtain the homograph information to get the camera matrix is a 3-D object matching. Finally, the proposed system to put the game on algorithm-based augmented reality, natural feature points of the game will be implemented.
목차
1. Introduction
2. Proposed Algorithm
2.1. Corner Location Information Extraction (Fast Corner Detector)
2.2. Feature Point Localization
2.3. I1I2I3 color Information Extraction
2.4. Extraction of the Size Direction of the Gradient
2.5. Limited Area Matching Method
3. Experiment Result and Verification
3.1. Algorithm Speed and Performance Result
3.2. Comparison with the Existing Algorithms
4. Conclusion
Acknowledgements
References