基于自适应梯度先验的旋转模糊图像复原算法

When the missile-borne infrared imaging system rotates with missile at a high speed rotation, a serious rotation blurring exists in the image acquired during the exposure, which causes a great confusion for subsequent target recognition and image tracking. A novel rotational blur restoration algorithm based on adaptive gradient priori regularization is proposed. In the proposed algorithm, the adaptive gradient priori regularization term is used for the deconvolution operation of the one-dimensional vectors extracted along the blurred fuzzy path from the image in the frequency domain. An adaptive median filtering algorithm using look-up table is designed to solve the problem about the hole points generated by the pixel extraction method using Bresenham algorithm. The experimentally simulated results show that, compared with improved Wiener filtering, constrained least square filtering, and gradient loading filtering, the proposed algorithm can effectively adapt to low SNR interference environment, and has strong ability to suppress noise and weaken the ringing-effect.