Active Contour Model for Image Segmentation
Hamza Zia, Asim Niaz, Kwang Nam Choi · 2022
Region based active contours algorithms are extensively utilised for image segmentation irrespective of unavailability of the densely annotated large data sets as required in the case of fully supervised deep learning models. However, previous active contours models have certain limitations including false contours appearances when there is in-homogeneity in the image. In our model we combine local and global information in image level set function, proposing a hybrid energy function which helps efficiently evolve contours on image and may assess the significance of the object and surroundings.Bias-correction is used it solve energy of the bias field that takes into consideration the intensity in-homogeneity and the level set functions that indicate a division of the image domain. The proposed model computes its data force using image fitting energy to take out local information from in-homogeneous image and calculates all pixel values by once. Objects having high contrast of different gray level value or more in-homogeneity can be segmented. Results shows that our method is more stable and take less computation time as compared to previous models. Finally the superiority of the proposed models in terms of segmentation efficiency is proved.