Y. Jiang (Canada)
Active contour model; prior shape matching
This paper describes a mathematical model of segmenting bone boundary from X-ray image that incorporates prior shape information into a geodesic active contour model. This model improved the level set method [1] in which shape information is missing from its energy function. The evolving process of the model in the level set stops when the prior shape and the evolving curve are aligned. The prior shape adjust its rotation, scale, and translation with the evolving curve by analyzing the displacement of their boundaries. The curve evolving process is constrainted by its prior shape. The difference between model and curve are measured by the boundary information. The algorithm is computation efficient of the order O(N). It improved the method by Chen[2] that aligns their model by the area calculation which the computation efficiency is O(N2 ). The experimental results are provided.
Important Links:
Go Back
