基于元胞自动机的动态背景运动目标检测

Moving target detection based on dynamic background of cellular automaton

针对传统运动目标检测算法在动态背景条件下难以准确检测出运动目标的问题,提出了一种基于元胞自动机的动态背景运动目标检测算法。首先,根据SLIC算法分割视频图像,并应用多模态混合动态纹理模型对视频图像进行背景建模。然后,融合空时显著性检测与基于元胞自动机的自动更新机制得到优化的显著性图。最后,通过对优化后的显著性图做适当的阈值分割处理得到视频图像中的运动目标。实验仿真结果表明,在动态背景条件下该算法可以有效的抑制视频图像中非运动目标的显著性物体对检测结果带来的影响,检测运动目标的精度较高,并且具有一定的鲁棒性。

Aiming at the problem that it is hard to use the traditional moving target detection algorithm to accurately detect the moving target under the dynamic background, a kind of moving target detection algorithm for the cellular automaton under the dynamic background was proposed in the thesis. Firstly, according to SLIC algorithm, video images were divided in the thesis, and multi-mode hybrid dynamic texture model was used for background modeling for video images; Then, space-time salience detection was integrated with the optimized salience map which was obtained based on the automatic updating mechanism of the cellular automaton; Finally, through making appropriate threshold segmentation process for the optimized salience map, moving targets in video images was obtained. The experimental simulation result shows that under dynamic background, the algorithm can effectively restrain the influence of the salient object for non moving targets in video images on the detection result; moving targets can be detected with higher accuracy; what＇s more, the algorithm has certain robustness.