CAJ | 학술논문

动态背景下运动图像的分割是一个重要问题。采用时空联合的分割方法，先对视频图像进行空域单帧分割，采用基于图论的分割技术，即找出一个包含最大面积的最小周长的边界；再根据运动图像间的变换信息，利用背景运动估计和补偿技术，得到全局运动参数。通过全局运动补偿，去除背景区域得到运动目标。实验证明，该方法有效地解决了过分割的问题，得到精确的分割效果。
동태배경하운동도상적분할시일개중요문제。채용시공연합적분할방법，선대시빈도상진행공역단정분할，채용기우도론적분할기술，즉조출일개포함최대면적적최소주장적변계；재근거운동도상간적변환신식，이용배경운동고계화보상기술，득도전국운동삼수。통과전국운동보상，거제배경구역득도운동목표。실험증명，해방법유효지해결료과분할적문제，득도정학적분할효과。
Moving image segmentation in dynamic background is an important issue.In this article we use spatiotemporal segmentation method,carry out the spatial domain single frame segmentation on video image first,which uses the graph theory-based segmentation technique,that is,to find out a boundary that enclosing the biggest area with minimum circumference;then we obtain the global motion parameters according to the variation information among the moving image and through making use of the background motion estimation and compensation technology.By global motion compensation,we remove the background region and procure the motion object.Practice proves that the method effectively resolves the over-segmentation problem and gets accurate segmentation effect.