An active contour model driven by K-means clustering for image segmentation

Active contour model (ACM) is one of the most often utilized approaches in the application of image segmentation owing to its excellent segmentation accuracy. However, common issues such as under-segmentation and over-segmentation may frequently take place by utilizing the majority of existing ACMs. To solve these issues, this study incorporates K-means (KM) clustering algorithm into the energy function, which can accurately process different kinds of images. In contrast with recently developed ACMs, the proposed KM model obtains better segmentation results (The CPU elapsed time T, iteration number N, and DSC). In addition, compared with fuzzy c-means (FCM) clustering algorithm, this model exhibits a relatively faster segmentation speed.