Absorbing Markov chain saliency driven active contour model for digital image boundary extraction / Muhammad Syukri Mazlin, Abdul Kadir Jumaat and Rohana Embong

The image segmentation approach using the active contour model (ACM) has achieved notable success in digital image analysis for extracting image boundaries. The region-based ACM can be divided into two classes: global segmentation and selective segmentation. Global segmentation, in which all desired boundaries of targeted objects are extracted from an input image, is preferable to the selective model due to its high utility, particularly in shape analysis. However, when segmenting a digital image with inhomogeneous intensity, the global ACM appears to produce unsatisfactory results. Thus, this research provides a new global ACM for segmenting digital images with an inhomogeneous intensity incorporating an absorption Markov chain (AMC) saliency image map and local image fitting principles. In addition, the Euler-Lagrange (EL) Partial Differential equation for solving the proposed model was provided. Thirty sets of digital images were used to validate the model. The accuracy of the proposed model, as indicated by Dice and Jaccard values, is approximately 3.81% and 10.63% higher, respectively, than that of the competing model, as determined by numerical analysis. In addition, the segmentation process taken by the proposed model is faster than the existing model. The proposed model has high potential to be extended into colour and multiphase formulations in future research.