지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술대회자료
- 저자정보
- 발행연도
- 2013.8
- 수록면
- 273 - 284 (12page)
이용수
초록· 키워드
오류제보하기
Owing to an increase in the number of trademark images, evaluating a new trademark’s distinctiveness has become a challenge. This paper proposes a new class of orthogonal moment functions based on the W-system of degree one and applies the proposed W-moments on trademark retrieval. W-system of degree one is a hybrid function system on interval [0,1] consisting of piecewise linear functions with multi-level jump discontinuities, and has great advantage in representing information with discontinuities and describing disjointed or articulated contour shape. The W-moment is easy to compute due to the concise expression of the basis functions of the Wsystem of degree one. Furthermore, the new moments eliminate the need for numerical approximations, since the basis set is orthogonal on the discrete domain of the image coordinate space. These properties make W-moments preferable to the conventional orthogonal moments, and can be used to accurately reconstruct the objects, especially, objects group. So the W-moments are suitable for describing complex shape consisting of multiple disjointed curves, such as trademark images. In this paper, the W-moment is applied on trademark retrieval. The W-moments are first derived from the trademark image, the similarity between query and trademark in the database is then evaluated using the Euclidean distance between their W-moment feature vectors, and the trademark retrieval is finally accomplished by comparing the similarities. The experiments are conducted on MPEG-7 CE2 and a self-constructed trademark data set to test the proposed moments. Comparison experiment results show that the new moments give preferable retrieval results and it is feasible and efficient in practical application.
목차
Abstract
1. Introduction
2. Calculation of the Shape Principal Orientation and Object Region
3. Brief introduction to W-system
4. The W-moments
5. Image feature description using W-moments
6. Trademark retrieval experiments
7. Conclusion
참고문헌 (0)
참고문헌 신청이 논문의 저자 정보
최근 본 자료
댓글(0)
0
UCI(KEPA) : I410-ECN-0101-2015-500-000938543