The paper introduces a new geometric model for active contours in image processing, based on a geometric partial differential equation (PDE). This model is intrinsically stable, satisfying the maximum principle, and allows for rigorous mathematical analysis. It enables the extraction of smooth shapes while avoiding sharp corners and angles. The model can also be adapted to find multiple contours simultaneously. Due to its stability, robust algorithms can be designed without parameters, making it suitable for practical applications. Numerical experiments are presented to validate the model's effectiveness. The classical method of snakes, which uses an energy functional to deform an initial contour towards the desired edge, is compared with the new geometric PDE model. The snakes method involves minimizing an energy functional that includes both internal and external terms. The internal energy ensures smoothness, while the external energy drives the snake towards salient image features. The new geometric PDE model, based on mean curvature motion, offers a more robust and theoretically sound approach to edge detection, addressing the challenges of smooth shape extraction, respect for singularities, and robustness in algorithm design.The paper introduces a new geometric model for active contours in image processing, based on a geometric partial differential equation (PDE). This model is intrinsically stable, satisfying the maximum principle, and allows for rigorous mathematical analysis. It enables the extraction of smooth shapes while avoiding sharp corners and angles. The model can also be adapted to find multiple contours simultaneously. Due to its stability, robust algorithms can be designed without parameters, making it suitable for practical applications. Numerical experiments are presented to validate the model's effectiveness. The classical method of snakes, which uses an energy functional to deform an initial contour towards the desired edge, is compared with the new geometric PDE model. The snakes method involves minimizing an energy functional that includes both internal and external terms. The internal energy ensures smoothness, while the external energy drives the snake towards salient image features. The new geometric PDE model, based on mean curvature motion, offers a more robust and theoretically sound approach to edge detection, addressing the challenges of smooth shape extraction, respect for singularities, and robustness in algorithm design.