基于小波分析的小型无人机目标显著性检测方法

Saliency detection method for small unmanned aerial vehicle based on wavelet analysis

在目标自动识别系统中,采用显著性检测方法可以迅速地将注意力集中在感兴趣目标上,同时忽略周围背景信息的干扰,实现目标的快速精确定位。为解决叶簇等复杂背景下所提取的显著区域存在边缘模糊和抗噪能力差的问题,提出一种基于小波分析的小型无人机目标显著性检测方法。通过对比CIE XYZ和Lab色彩空间模型与不同小波函数对检测结果的影响,利用小波逆变换在X、Y、Z分量上分别重构产生多幅特征图,并应用sym2小波进行显著图解算,论述了显著性检测方法的性能评价指标。实验结果表明,sym2小波能够实现叶簇背景下无人机目标显著区域的有效检测。

Saliency detection methods are used in automatic targets recognition system can quickly focus on interest object and ignore the interference of surrounding background,to locate targets fast and accurately.A saliency detection method for small unmanned aerial vehicle based on wavelet analysis is presented,to solve the edge blur and poor antinoise ability of saliency region extracted from complex background such as leafage.Effects of saliency detection results in CIE XYZ and Lab color space,and different wavelet functions are contrasted.Multiple feature maps are created on X、Y、Zcomponents in reconstruction process using inverse wavelet transform,and sym2 wavelet is adopted to calculate saliency maps.The performance evaluation indexes of saliency detection method are discussed.The result shows sym2 wavelet can effectively detect the saliency region of unmanned aerial vehicle in leaf cluster background.