A geometric active contour model without re-initialization for color images

A geometric active contour model without re-initialization for color images is proposed in this paper. It combines directional information about edge location based on local squared contrast as a part of driving force, together with the improved geodesic active contour containing Bayes error based statistical region information as well as an extra term that penalizes deviation of the level set function from a signed distance function. All these measures are integrated in a unified frame thus the costly re-initialization procedure can be completely eliminated. Experimental results on real color images have shown that our model can extract contours of objects in images precisely and its performance is much better than the Geodesic-Aided C-V (GACV) model.