作者: Chi-Shiang Chan , Yung-Kuan Chan , Cheng-Ping His , Chun-Chieh Wang
DOI: 10.1109/ICCSN.2011.6014754
关键词:
摘要: Object contour detection plays an important role in image processing. Many edge and object detectors have been proposed, such as, Canny detector, ACM, level set methods. However, those methods often cannot provide the fully perfectly. In ACM methods, initial of has to be specified by a user advance, successful segmentation results highly depend on given contour. addition, are significantly sensitive noise. Giving is not required for but it may bring about "broken problem." To solve broken problem, feature-based disconnected segments (FBDES) linker provided this paper. For extracting objects from image, paper detector first used detect contour, then FBDES employed link based features gray-level difference at vicinity of, gradient direction on, length line segment connecting two segments. The experimental show that can give more impressive than method.