基于区域融合的小目标图像盲复原方法

Blind Restoration Method of Small Target Image Based on Region Fusion

针对高超声速飞行器的复杂光学成像条件下出现的图像模糊与噪声问题,提出一种基于区域融合估计点扩散函数的小目标图像盲复原方法,从目标区域提取和区域融合两个方面降低小目标图像点扩散函数估计误差。首先利用边界延拓解决目标与边界的重叠问题,然后使用多区域加权融合结合灰度匹配的方法,提高点扩散函数的估计准确性。采用仿真数据和实际图像验证所提出方法的有效性。结果表明:该方法可以有效抑制背景噪声,集中目标的能量分布,改善小目标的清晰度和显著性。本研究为高超声速飞行器的小目标识别与精确打击奠定了基础。

This paper has proposed a blind image restoration method using region fusion for estimating the point spread function to address the blur and noise in complex optical imaging conditions of hypersonic vehicles.Extracting the target region and fusing multiple regions can reduce the estimation error of the point spread function of small target images.Firstly,the boundary extension was utilized to resolve the overlap between the target and image boundary.Then,the estimation accuracy of the point spread function was optimized using multi-region weighted fusion and grayscale matching.After that,the effectiveness of the proposed method has been validated by simulations and real images.The results show that this method can effectively suppress background noise and concentrate the energy distribution of the target,thus improving the sharpness and saliency of small targets.This study provides the foundation for small target recognition and precision strikes of hypersonic vehicles.