一种目标兴趣区的快速识别算法

A FAST RECOGNITION ALGORITHM FOR INTERESTING REGION

本文给出了一种基于角特征的目标兴趣区的识别算法。通过对二值图象边缘点的向前和向后K步跟踪,可以得到该点两个方向的方向链码,而一个角则由该两个方向链码确定。然后根据已知目标上角分布的先验信息,将目标的主体提取出来。本文给出的算法可并行实现。计算机模拟结果表明,该算法是有效的。

A fast recognition algorithm is presented for interesting region extraction of an object based on corner features. Two direction code chains of boundary elements are obtained through forward and backward boundary tracking, and a corner is determined by the code chains. The dominate part of the object can be extracted by using prior information of corner distribution on the digital image. The algorithm suggested herein can be performed on a parallel basis. The simulation results proved it to be effective.