한국정보처리학회논문지 (The Transactions of the Korea Information Processing Society)
- 제4권7호
- /
- Pages.1865-1872
- /
- 1997
- /
- 1226-9190(pISSN)
화상 구조 파악에 의한 화상의 잡음 제거 및 경계선 추출
Noise Removal and Edge Detection of Image by Image Structure Understanding
초록
본 논문은 기존 경계 추출 방법의 문제점인 임계치 선정 문제와 평활화로 말미암아 경계에서 발생할 수 있는 흐림 현상을 제거할 수 있는 방법을 제안함을 목적으로 한다. 이를 위해
This paper proposes not only the thresholding problem which has been one of the major problems in the pre-existing edge detection method but also the removal of blurring effect occurred at the edge regions due to the smoothing process. The structure of a given image is assigned as one of the three predefined image structure classes by evaluating its toll membership value to each reference structure class:The structure of an image belongs to the structure class which has the least cost value with the image. Upon the structure class assigned, noise removal and edge extraction precesses are performed, e.g., the smoothing algorithm is applied to the image if its structure belongs to the pure noise region class; edge extraction while removing the noise is performed simultaneously if the edge structure class. The proposed method shows that preventing the blurring effect can be usually seen in the edge images and extracting the edges with no using thresholding value by the experiments.
키워드