基于方向选择性机制的图像背景杂波建模方法

Image background clutter modeling method based on directional selectivity mechanism

受视觉系统方向选择性机制启发,提出了种新的图像杂波建模方法,并将其应用于成像系统目标获取性能评估中。首先,利用像素的梯度方向模拟视觉系统局部感受野中神经元的响应方向,通过比较中心像素与邻域像素之间的梯度方向相似性模拟神经元“兴奋”与“抑制”响应之间的关系,设计了一种方向选择性视觉模式。其次,考虑到人眼视觉系统对亮度对比度敏感,将亮度对比度作为权重因子对视觉模式直方图加权,在加权直方图空间中对背景杂波建模。最后,推导了背景杂波与目标获取性能之间的关系,建立了光电成像系统目标获取性能评估模型。利用公开数据集对模型进行了验证,实验结果表明,提出的方法所评估的目标获取性能与外场实际目标获取性能具有较高的一致性,而且在均方根误差、相关性方面优于现有的性能评估方法。

Inspired by the direction selectivity mechanism of vision system, a new image clutter modeling method was proposed and applied to the target acquisition performance evaluation of imaging system. Firstly, the gradient direction of pixels was used to simulate the response direction of neurons in the local receptive field of the visual system. By comparing the gradient direction similarity between the central pixel and the adjacent pixel,the relationship between neuron "excitation" and "inhibition" response was simulated. The visual pattern was designed based on directional selectivity. Secondly, considering the sensitivity of human visual system to brightness contrast, the brightness contrast was used as a factor to weight the visual pattern histogram. The background clutter was modeled in the weighted histogram space. Finally, the relationship between background clutter and target acquisition performance was deduced, and the target acquisition performance evaluation model of photoelectric imaging system was established. The experimental results show that the target acquisition performance evaluated by the proposed method is consistent with the actual target acquisition performance in the field, and is superior to the existing performance evaluation methods in terms of root mean square error and correlation.