Controlling B-Spline Snake Behavior using Particle Swarm Optimization

Traditional active contour models are hardly convergence to object boundaries where subjective contour and complex background appear. One of the main reasons is because the weighted combination of image energy functions can produce totally different B-Spline snake behaviors. Some efforts have been focused to improve the external energy forces but also lead to higher computational cost. In this research, an approach based on the particle swarm optimization is utilized to control the snake behavior without the necessity of computing the additional information. In the proposed approach, the weighted parameters are optimized using particle swarm optimization by evaluating control points of B-Spline snake in the image energy terms. This method enables the snake to initialize the image energy coefficient automatically and in more convenient way. Experimental results demonstrate that the proposed method can improve the snake behavior.