湍流退化图像的最大熵盲目复原方法

Blind restoration of turbulence-degraded image using maximum entropy algorithm

装有成像制导装置的飞行器在大气中高速飞行时由于受湍流的影响,接收到的图像发生严重退化,大大影响制导精度。针对该问题,提出了一种针对具有单一背景的湍流退化图像的最大熵盲目复原方法。复原过程中,为避免最大熵约束项的非线性带来的复杂运算,对熵的表达式进行二次近似,并利用灰度变换确保近似程度的准确性,最后采用共轭梯度法进行求解,从而大大降低了计算量。对不同马赫数下得到的湍流退化图像进行试验,验证了该方法的有效性。

When the high-speed aircrafts with imaging guidance system fly in the atmosphere, the received images are greatly degraded due to the atmospheric turbulence. Thus the guidance precision is greatly constrained. A new maximum entropy blind restoration method to improve the degraded images having single background was proposed. The nonlinear entropy expression was approximated by the two order polynomial to reduce the complex computation caused by the nonlinearity of maximum entropy regularized terms. Theoretical analysis shows the accuracy of the approximation can be assured by the density transformation. The problem can be solved by the conjugate method wl-fich needs few computation. Then series of turbulent degraded images obtained from different conditions are restored to evaluate the proposed method.