원문정보
보안공학연구지원센터(IJMUE)
International Journal of Multimedia and Ubiquitous Engineering
Vol.9 No.1
2014.01
pp.403-414
피인용수 : 0건 (자료제공 : 네이버학술정보)
초록
영어
In this paper, we propose an intelligent method to extract flaws from nondestructive testing(NDT) images of ceramics. Our goal is to extend the previous study[8]'s feasibility to handle images having skewed brightness distribution or darker images. Thus, we use Gaussian filtering along with anisotropic filtering and sigma fuzzy binarization instead of repetitive binarization used in [8]. These techniques enable us to enhance the brightness contrast successfully so that the usability of our method is extended and verified in experiment. Two phases of image processing - area segmentation and extraction of the defect by searching for labeled pixel to form a defected object by Grassfire algorithm - are explained in detail.
목차
Abstract
1. Introduction
2. Area Segmentation by Slope Analysis
2.1. Image Transform by Ends-in Search Stretching
2.2. Removing Minute Noise by Anisotropic Mask and Gaussian Filtering
2.3. Extracting Boundary Line with 7 X 7 Sobel Mask
2.4. Area Segmentation by Slope Analysis of Boundaries
3. Extracting the Defect
3.1. Area Segmentation by Grassfire labeling
3.2. Extract the Defect by Sigma Fuzzy Binarization
3.3 Removing Noise with Morphological Information
4. Experiment and Analysis
5. Conclusion
References
1. Introduction
2. Area Segmentation by Slope Analysis
2.1. Image Transform by Ends-in Search Stretching
2.2. Removing Minute Noise by Anisotropic Mask and Gaussian Filtering
2.3. Extracting Boundary Line with 7 X 7 Sobel Mask
2.4. Area Segmentation by Slope Analysis of Boundaries
3. Extracting the Defect
3.1. Area Segmentation by Grassfire labeling
3.2. Extract the Defect by Sigma Fuzzy Binarization
3.3 Removing Noise with Morphological Information
4. Experiment and Analysis
5. Conclusion
References
저자정보
참고문헌
자료제공 : 네이버학술정보