Abstract: As cubic spline curve generated by the manually selected point can not describe the image edge accurately, the current image edge description algorithm is improved by using the particle swarm cubic spline optimization algorithm. The value points will be optimized by the particle swarm optimization and then be used to generate a spline curve. Experiments show that the algorithm can generate the spline curve faster, and the spline curve generated by this algorithm describes the image edge more accurately.

Key words: cubic spline, image edge, edge expression, Particle Swarm Optimization（PSO）, edge detection operator

摘要： 针对手工选点生成的三次样条曲线无法准确地描述图像边缘的问题，以粒子群三次参数样条曲线优化算法为基础对现有的图像边缘算法进行了改进。在图像边缘上取若干点作为样条曲线的型值点，利用粒子群算法对这些型值点进行优化，使得到的样条曲线对图像边缘有很好的贴合。实验表明，该算法不仅能够快速拟合样条曲线，而且优化后的样条曲线形态能较好地贴合图像边缘轮廓。

关键词: 样条曲线, 图像边缘, 边缘表达式, 粒子群优化, 边缘检测算子