CAJ | 학술논문

针对运动背景下运动目标分割定位困难的问题，提出了融合背景补偿与均值漂移的运动背景下目标分割定位算法。该算法首先在运动背景补偿的基础上，利用帧间差分法得到差分图像。其次，对差分图像中非零像素点建立多特征描述子，运用均值漂移算法对其进行聚类。最后，利用均值聚类得到的非零像素点来对运动目标进行分割定位。实验结果充分表明，该算法可以比较精确地进行运动背景下移动目标的分割定位。
침대운동배경하운동목표분할정위곤난적문제，제출료융합배경보상여균치표이적운동배경하목표분할정위산법。해산법수선재운동배경보상적기출상，이용정간차분법득도차분도상。기차，대차분도상중비령상소점건립다특정묘술자，운용균치표이산법대기진행취류。최후，이용균치취류득도적비령상소점래대운동목표진행분할정위。실험결과충분표명，해산법가이비교정학지진행운동배경하이동목표적분할정위。
In order to solve the problem of segmentation and location of moving object in moving scenes, an algorithm of segmentation and location of moving object in moving scenes with the fusion of background compensation and mean shift is proposed. Firstly, frame-difference method is used to get difference image based on moving scenes compensation. Then, multiple feature descriptors of nonzero pixels in difference image are established and mean shift algorithm is adopted to deal with the nonzero pixels. Lastly, segmentation and location of moving object is done with the nonzero pixels gotten from mean shift clustering. The results indicate that this method can realize moving object detection accurately.