【摘要】 针对无特定背景条件下运动目标的分割问题,采用改进的三步搜索算法有效地提高灰度投影法检测序列帧间运动矢量搜索速度,同时保持了运动矢量的搜索精度。并通过帧间运动矢量补偿将相邻三帧图像的相同背景稳定在同一幅图像的相同位上,提出了改进型HSI帧差分法检测运动目标区域,通过聚类分析确定运动目标的个数。结合数学形态学的闭运算和填充运算分割出运动目标。实验结果表明,该方法能够快速高效地从无特定背景条件下的视频序列中提取运动目标,具有图像信息利用充分、鲁棒性强和准确性高等优点。更多还原

【Abstract】 For the problem of segmenting moving objects in unspecific background condition,in this paper we adopted improved three-step searching algorithm to effectively accelerate the searching speed of gray projection algorithm detection on motion vectors in sequence frame,while keep up searching accuracy on motion vectors.The same backgrounds of adjacent three frames are fixed to the same position of the same frame through intra-frame motion vector compensation.Then we proposed to use improved HIS frame difference algorithm to detect moving object’s area,and to determined the number of moving objects with clustering analysis.Finally,the moving objects were segmented in combination with closing algorithm and filling algorithm of the mathematical morphology.Experimental results demonstrated that this method can effectively and quickly extract moving objects from video sequence images in unspecific background condition,and has the advantages of strong robustness,high accuracy and sufficient making use of the image’s information.更多还原