摘要
海浪由于在形成过程中受到物理规律的制约,从而呈现与海岸线近似于平行的弱边缘。相反的,物体则在各个方向上表现出比较强的边缘性。根据海浪和物体的这种不同成像特性,提出在单帧图像中采用Sobel算子计算梯度图像从而分割出感兴趣区域,然后从感兴趣区域出发,根据连续帧间的感兴趣区域的中心点是否移动来判断物体是否运动,从而找出运动的目标。实验表明,该方法鲁棒性好,能有效地减小误检和漏检的概率,较准确地检测到海上的运动目标。
Because of being restricted by the physics rule during the forming process,ocean wave presents feebleness fringe approximately paralleled to the coasting.Contrary,objects present stronger fringe in all directions.Based on the different imaging characteristic between ocean wave and objects,this paper brings forward segmenting those regions of interest in one frame of image using Sobel arithmetic operators,and then detecting the moving targets in those regions of interest based on the moving of the center of the interested region.Experimental results show that the method can minish the probability of dis-detecting and leak-detecting,and detect the moving targets robustly and accurately.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第4期225-226,231,共3页
Computer Engineering and Applications
关键词
成像特性
SOBEL算子
感兴趣区域
中心点
imaging characteristic
sobel arithmetic operators
regions of interest
center
