摘要
提出了一种基于图像对称差分运算的运动小目标检测新方法．该方法以连续三帧序列图像为一组处理对象，先进行绝对差运算，再作均值滤波，保证了算法能很好地检测出复杂背景中的低信噪比、运动小目标；单帧二值图像处理以及基于运动轨迹连续性的目标检测后处理，可以在保证系统检测概率的前提下减少目标检测的虚警概率．实验表明，该方法性能比图像差分法要好，与累积图像差分法相差无几，且硬件实现容易．
A new detecting method is studied for a small moving target based on symmetry difference operations. The operations, with absolute difference and average filter, are undertaken in the three frames of image sequences to guarantee the efficiency of detecting moving target of small size with low SNR from the complex scene. After processing the single binary image and detecting the target based on the moving continuity, the method can diminish false alarming targets under a constant probability of the detection. The method is implemented easily in hardware. The experimental results verify the validity of the algorithm compared with the operations of single difference or accumulating difference.
出处
《华中理工大学学报》
CSCD
北大核心
1998年第9期34-35,38,共3页
Journal of Huazhong University of Science and Technology
