基于超效率数据包络分析模型的数码迷彩融合特性评价方法研究

针对活动目标的数码迷彩伪装，为了从光学波段方面评价多种数码迷彩在不同背景中的融合性，对图像内容进行分析，选用HSV颜色模型，分别从图像的色调（H）、饱和度（S）、亮度（V）3个分量 提取数码迷彩和背景图像的颜色、纹理、形状、熵及复杂度等5个方面的特征值。根据超效率数据包络分析模型原理，把数码迷彩与背景图像中各个特征差异值作为决策单元（DMU）的输入量，计算出迷彩与背景一对一融合性的效率值。将效率值数据取倒数作为新的DMU输入量，计算得到每种迷彩对不同背景融合性的效率值，从而实现数码迷彩对背景图像综合适应性的评价。利用图像显著性指标对评价结果进行验证，将融合性效率值最大与最小的两种数码迷彩对背景实施伪装，前者在背景图像中的显著性低，具有更好的背景融合性。

An Evaluation Method for Fusion Feature of Digital Pattern Painting Based on Super-efficiency DEA Model

To evaluate the fusion feature of various digital pattern painting in different backgrounds for the camouflage painting of moving target,the optical images are analyzed,the hue-saturation-value (HSV)model is used,and the image features about color,texture,shape,entropy and complexity are extracted from digital pattern painting and background images.On the basis of super-efficient data envelopment analysis (DEA) model,the efficiency values of one-to-one fusion of digital pattern painting and background images are calculated by taking the differences among the above features in digital pattern painting and background images as the inputs of decision-making unit (DMU).And the reciprocal efficiency values are used as the inputs of new DMU unit,and the efficiency values of fusion of digital pattern painting and background are calculated for the evaluation of comprehensive adaptability of digital pattern painting to background.The efficiency values are checked using the saliency index of image,and two digital pattern painting with maximum and minimum efficiency values are used to camouflage the background.The result shows that the former has lower saliency and better integration in background.