Title :
A novel non-local geodesic active contour model
Author :
Qiong Lou ; Dexing Kong
Author_Institution :
Dept. of Math., Zhejiang Univ., Hangzhou, China
Abstract :
In this paper we propose a novel variational model by combining the non-local information and the geodesic active contour model, and make use of the non-local information between patches in the neighborhood of interested pixel which is large enough. We define a non-local energy in the sense of intensity by summing up the interactions between pairs of patches that may have similar features, where patches with close distance and similar intensity may have larger non-local values. Adjoint with geodesic active contour, we make use of the advantage of edge-detector and region features. In our model, because of patch information, we only need a local homogeneity of features, thus it is less sensitive to noise and initialization than geodesic active contour model.
Keywords :
edge detection; image segmentation; object detection; edge detector; image segmentation; level set method; novel non local geodesic active contour model; object detection; patch information; region features; Active contours; Adaptation models; Computational modeling; Image segmentation; Level set; Mathematical model; Measurement;
Conference_Titel :
Image and Signal Processing (CISP), 2012 5th International Congress on
Conference_Location :
Chongqing
Print_ISBN :
978-1-4673-0965-3
DOI :
10.1109/CISP.2012.6469962