论文部分内容阅读
提出了一种封闭式图像盲分离算法,用以解决当前盲图像处理问题.利用观测的混合图像二次特征函数二次导数矩阵与其对角分量和的二次导数矩阵联合近似对角化,估计未知的混合矩阵,进而用混合矩阵的逆阵与观测矩阵的乘积恢复原始图像,实现图像的盲分离.本算法无需数据存储和迭代,避免了采用高阶统计量的严重运算负担,分离效果显著,试验仿真结果验证了算法的有效性.
A blind image blind separation algorithm is proposed to solve the problem of current blind image processing.Using the observed approximation of the quadratic derivative matrix of the quadratic eigenfunction of the mixed image and the second derivative matrix of the diagonal components and the diagonal component, Estimate the unknown mixture matrix and then restore the original image by using the product of the inverse matrix of the mixed matrix and the observation matrix to realize the blind separation of the image.This algorithm avoids the need of data storage and iteration and avoids the heavy computational burden of using higher order statistics, Significantly, the experimental simulation results verify the effectiveness of the algorithm.