In order to classify odors that are in the air at the specific place and at the specific period, it is necessary to generate representative patterns for each odor sources. In this study, in order to classify the various odors efficiently, the k-mean clustering algorithm was applied to the experimental data. Considering the mixed air with various odors, the representative patterns for odor sources were generated with the combination of two or three different odors. As a result, an acquired odor sample could be discriminated and hence its source can be traced by the proposed clustering method, in this study. The results of this study will be useful for settling the civil complaint related odor.

ABSTRACT
요약문
1. 서론
2. 연구방법
3. 실험내용 및 결과
4. 결론
사사
참고문헌

• 유숙현
• 박상진
• 이희성
• 구윤서
• 권희용