启发策略的确定性退火技术非刚性点匹配算法

Non-rigid Point Matching Arithmetic Based on Deterministic Annealing with Heuristic Strategy

点匹配问题一直是计算机视觉、模式识别、医学临床诊断领域的一项重要的基础性工作。提出了带启发策略的确定性退火技术非刚性点匹配算法。该算法引入软匹配思想将匹配矩阵系数连续化,为加快匹配矩阵的收敛速度,给出了带启发策略的对应启发矩阵,构造了一种适合确定性退火算法求解的新能量函数公式。实验结果表明,该算法可以得到稳定的解,并且可以避免出格点等噪音的干扰,具有稳定性,有效性,和较高的效率。

Point matching plays an important role in computer vision,pattern recognition and medical image analysis. This paper presents a novel non-rigid point matching arithmetic with heuristic strategy based on deterministic annealing. In this algorithm,combinatorial optimum is transferred to consecutive optimum in computing the match-matrlx.The heuristic matrix with heuristic strategy corresponding to the match-matrix is provided in order to accelerate the convergence of matching,and an especial energy function suited for deterministic annealing is constructed in this algorithm.Experiments results illustrate that this arithmetic can reach a steady performance of matching,and can well avoid the disturbance of the anomalous points.It is a robust and effective arithmetic for non-rigid point matching.