A Gaussian-Extremum Image Pre-fitted Active Contour for Accurate and Faster Image Segmentation

Active contour models are techniques employed in image processing to achieve tasks such as image segmentation and object tracking. To optimize the process of segmentation and achieve better results, various hybrid models have been derived where characteristics of multiple models are utilized. In this work, one such model named Optimized Local Pre-Fitting Image energy (OLPFI) has been chosen for modification, inspired by the model Local Pre-Piecewise Fitting Bias Corrections (LPPF BC). The aim is to control the evolution of the contours to increase speed and to reduce noise tolerance effectively by applying selective weighting to the neighboring pixels. The introduction of the Gaussian average in the pre-fitting functions of this model has proved to be quite beneficial in terms of gaining better localization and handling of noise. Therefore, the proposed model gives higher accuracy along with improved time efficiency and noise tolerance.