Due to the frequency of occlusion, cluttering and low contrast edges, gray intensity based active contour models often fail to segment meaningful objects. Prior shape information is usually utilized to segment desirable objects. A parametric shape prior model is proposed. Firstly, principal component analysis is employed to train object shape and transformation is added to shape representation. Then the energy function is constructed through a combination of shape prior energy, gray intensity energy and shape constraint energy of the kernel density function. The object boundary extraction process is converted into the parameters solving process of object shape. Besides, two new shape prior energy functions are defined when desirable objects are occluded by other objects or some parts of them are missing. Finally, an alternating decent iteration solving scheme is proposed for numerical implementation. Experiments on synthetic and real images demonstrate the robustness and accuracy of the proposed method.
展开▼
