摘要
为了有效检测出海空背景的舰船目标,给出了一种扩展的元胞自动机(CA)的δ参数求取方法,并针对海空目标图像的特点,提出了一种基于CA混沌特征提取的海空目标检测方法。首先对连续的多帧视频图像序列进行分块,然后利用CA的δ参数提取图像序列的混沌特征,标记没有混沌特征的图像区域,最后依据种子点和区域生长法确定目标区域。在连续的100帧视频图像序列上进行仿真实验,结果表明新的方法检测率更高,优于灰度统计和谱分析方法。
For detecting efficiently navy-ship target in sea-area background, it proposed a new algorithm of computing extended parameter B of CA, and according to images＇ feature of sea-area, a new sea-area target detection algorithm based on chaos feature extraction of CA was proposed. First, block several frame continuous video images; then, extract chaos feature of image sequence according to parameter δof CA, and marked the area which have no chaos feature; finally, the target area was detected according to seed points and the method of regional growth. With an experiment on 100-frame continuous video images, the result shows that the detection ratio this way is higher than gray computing and spectral analysis.
出处
《电视技术》
北大核心
2010年第11期104-106,113,共4页
Video Engineering
基金
中国博士后科学基金项目(20100471451)
