@article{oai:bunkyo.repo.nii.ac.jp:00003190,
 author = {Shimodaira, Hisashi},
 journal = {情報研究, Information and Communication Studies},
 month = {2011-01-01, 2011-01-25},
 note = {In an image, boundaries between regions represent the shapes of objects in the scene. Finding the boundaries　is　a　problem　of　fundamental　importance　in　the　area　of　computer　vision　such　as　object　recognition whose accuracy depends　on　the　quality　of　the　estimates of　the　boundaries.　This paper　proposes　a　method　for　detecting　boundary edges based on a local image-feature integration method using a modified Canny edge detector. In the proposed method, first by edge saliency indices , salient edge are selected from the output of the modified Canny edge detector. Next , by boundary likelihood indices, edges with high boundary likelihood are selected from the salient edges. The major features of  our method are as follows ; the algorithm is simple and easy to implement ; the computational complexity is not so high ; the paramater values once tuned for several images can be used in common for other images ; the good thresholds for selecting boundary edges are automatically computed. We have performed experiments on various natural images. The results have shown that our method works effectively and can produce relatively good estimates of boundary edges and its performance is robust.},
 pages = {15--40},
 title = {A Method for Detecting Boundary Edges Based on a Local Image-Feature Integration Method},
 volume = {44},
 year = {}
}