基于分形特征的复杂环境目标检测方法研究

A Research of Moving Target Detection under Complex Background Based on Fractal Feature

摘要:

针对复杂的自然背景下的运动目标检测,提出了一种基于分形特征的运动目标检测算法；该算法利用目标的分形维数与自然背景分形维数的差异将目标从背景检测出来；首先应用改进的地毯覆盖法快速得到图像的分形维数,然后通过比较邻域之间分形维数的相互关系进行目标检测；实验结果表明,该方法能对复杂背景下的运动目标进行检测,由于采用分块求分形特征的方法,能有效地减少搜索目标所带来的计算量,算法过程简单、检测速度快、检测结果精确,目标与背景对比度的变化对检测结果几乎没有影响, 且噪声对该算法的检测结果影响较小；在运动目标实时检测问题上有着很好的实用价值和应用前景。

Abstract:

A moving target detection algorithm based on fractal feature is proposed, which is suitable for complicated nature background. First, the improved carpet cover method was used to get the fractal dimension of picture quickly. Second, the target was detected through the differences of fractal dimension between neighborhood. Experimental results show that the algorithm can detect a moving target under complex background, and has fast speed and accurate result, and this algorithm has good practical value and application prospect on moving target real-time detection.