<p>The active contour model has been a widely used method in image segmentation that remains challenging due to the noise and intensity inhomogeneity. In this paper, an adaptive fractional-order edge-stopping function-based active contour model is proposed. To better preserve the weak texture information in the image, a new adaptive fractional order matrix, defined using the gradient information of the image, is first constructed. Then, a new edge-stopping function based on adaptive fractional differentiation is utilized to improve the robustness of the proposed model to noise. Finally, the energy fitting term based on regional variance can achieve an arbitrary selection of the initial position of the active contour curve. Experimental results on grayscale and natural images in PASCAL VOC2012 demonstrate that this method can effectively segment images with noise and intensity inhomogeneity. Compared to the optimal competitive active contour model WHRSPF, the proposed method shows an average increase of 7.9% and 4.5% in IOU and DSC, respectively.</p>