CAJ | 학술논문

针对传统轮廓跟踪算法易出现漏跟踪和跟踪断裂的缺点，提出交叉点的概念。利用交叉点结构完整地描述边界跟踪的不同方向，以保证每一个方向都能被跟踪到。在此基础上，形成以交叉点为结点的目标边界的树结构描述。最后通过对树的遍历获得目标边界的跟踪序列。实验结果表明该算法速度快，边界识别准确，尤其适于识别大型复杂目标的轮廓。
침대전통륜곽근종산법역출현루근종화근종단렬적결점，제출교차점적개념。이용교차점결구완정지묘술변계근종적불동방향，이보증매일개방향도능피근종도。재차기출상，형성이교차점위결점적목표변계적수결구묘술。최후통과대수적편력획득목표변계적근종서렬。실험결과표명해산법속도쾌，변계식별준학，우기괄우식별대형복잡목표적륜곽。
In the paper we present the concept of cross-point to resolve the problem of overlooked tracking and incomplete tracking easily happened in traditional boundary tracking algorithm.The structure of cross-point is used to completely describe the different directions of boundary tracking so as to ensure that every direction can be tracked.On this basis,the tree structure description of the target boundary which takes the cross-point as the node is made up.Finally the tracking sequence of target boundary is obtained by traversing the tree.Experimental results show that the algorithm has high speed and accurate boundary identification;it is especially suitable for recognising the large and complicated boundary.