CAJ | 학술논문

传统算法通常需在头部稳定或光线明亮的情况下进行，较难在自然条件下对多姿态头部运动时的人眼进行跟踪。针对该问题，提出基于Kinect传感器的瞳孔定位算法，使用三维主动表观模型（ AAM）对眼轮廓匹配得到眼部特征点，粗定位分割出眼睛，再对瞳孔进行精确定位。实验表明：在多姿态与头部带有遮挡物的情况下仍能较好地跟踪和定位人眼。
전통산법통상수재두부은정혹광선명량적정황하진행，교난재자연조건하대다자태두부운동시적인안진행근종。침대해문제，제출기우Kinect전감기적동공정위산법，사용삼유주동표관모형（ AAM）대안륜곽필배득도안부특정점，조정위분할출안정，재대동공진행정학정위。실험표명：재다자태여두부대유차당물적정황하잉능교호지근종화정위인안。
Traditional pupil localization algorithm can only be applied while head is stable or lighting condition is bright,so it is difficult to track human eye in multipose head movement. Aiming at this problem,propose pupil localization algorithm based on Kinect sensor,establish a 3D active appearance model(AAM)to match bounding of eyes to obtain feature points,coarse positioning is carried out to segment eye,then pupil accurate positioning is carried out. Experimental results show that the proposed method can still track and locate human eye in condition of multipose and head is sheltered.