摘要
探讨一种基于双目视觉的呼吸运动实时跟踪方法,以减少在胸腹部的肿瘤放射治疗中由于呼吸等因素造成肿瘤位置动态位移而引起的治疗误差。使用由双摄像机组成的计算机视觉系统,实时匹配出标记物在左右两摄像机采集的图像中的具体坐标,依据双目成像的基本原理,计算出标记物在腹部表面的三维坐标值,结合时间参数,计算出该特征点三维坐标的变化情况,以此来完成对呼吸运动的实时跟踪。在目标跟踪过程中,使用鲁棒性强的SIFT(scale invariant feature transform)算法作为目标图像匹配的方法,并且在算法设计过程中,采用动态选择待匹配图像和局部搜索的策略。实验结果表明,目标图像匹配精确,减少了提取图像SIFT特征所需要的大量时间。在一个呼吸周期内,视觉测量的标记物的最大运动范围与实际测量值不超过0.2 cm,且能够做到实时性计算。该模型的运行精度高,能够较好地实时跟踪人体的呼吸运动。
In radiotherapy,tumors and surrounding tissues locating inside the thorax or abdomen are deformed in cycles with the physiological or physical influence such as breathing and heartbeat.In order to reduce the influence,we developed a binocular stereo vision based real-time tracking for respiratory motion.The respiratory movement was tracked by two cameras composed of computer vision system,which is real-time matching a marker′s coordinates in the two image acquisition by the cameras that was placed in the abdomen surface.According to the basic principle of binocular,three-dimensional coordinates of the marker was calculated.Combining time parameters the variation of 3D feature points was calculated to track the respiratory movement in a real time manner.SITF(scale invariant feature transform) was used as the registration algorithm,which had strong robustness and accurate matching performance in tracking process.In order to reduce time overhead,dynamic choice matching image and local searching strategy were used in calculation and mergence.Experimental results showed that in a breathing cycle,the influence of the biggest sports scope was not larger than 0.2 cm between the vision measuring and practical measurement,and can achieve real-time calculation.The proposed method is characterized by good robustness and effectively achieves real-time multi-target tracking in complex occlusion scenes for respiratory movement tracking.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2011年第4期520-527,共8页
Chinese Journal of Biomedical Engineering
基金
国家重点基础研究发展(973)计划(2010CB732500)
广东省科技计划项目(201031100013)
关键词
放射治疗
呼吸运动
图像跟踪
双目视觉
尺度不变特征变换(SIFT)
radiotherapy
respiratory movement
image tracking
binocular stereo vision
scale invariant feature transform(SIFT)