A Fractional Order Derivative Based Active Contour Model for Simultaneous Image Despeckling and Segmentation

Image segmentation faces significant challenges due to the presence of intensity heterogeneity and noise in real-world images. Specifically, medical ultrasound images are usually corrupted by speckle noise and intensity heterogeneity. To address these challenges, we introduce a new level set-based variational model by utilizing fractional edge terms. Our model comprises two equations, one performing image segmentation and the second suppressing the speckle noise simultaneously. The initial contour is guided towards the target boundaries by the segmentation equation, resulting in accurate segmentation. Simultaneously, the despeckling equation diminishes the influence of speckle noise, thereby improving the quality of segmentation. We perform comparison experiments on natural and medical images to demonstrate the efficiency of the present fractional active contour model (FACM). These images, especially ultrasound images, are characterized by speckle noise and intensity heterogeneity. Several segmentation measures are utilized to evaluate the performance of our present model. Through experimental outcomes, we demonstrate that our model surpasses most existing active contour models thus providing superior segmentation outcomes.