基于自适应增益系数的两级反锐化掩模法

Two-Stage Unsharp Masking Algorithm Based on Adaptive Gain Coefficient

传统的反锐化掩模法(Unsharp Masking Algorithm, UMA)采用固定的增益系数,难以在抑制噪声放大和增强图像特征之间实现较好的平衡.因此,提出了一种基于自适应增益系数的两级UMA.首先,采用UMA重点增强图像的细节;然后,采用加权核范数最小化(Weighted Nuclear Norm Minimization, WNNM)算法去除噪声;最后,再次采用UMA重点增强图像的强边缘.其中,两次UMA均采用基于梯度模的增益函数,自适应地调整细节处和强边缘处的增益系数.实验结果表明,相比传统的UMA,所提方法在增强工业X射线图像重要结构特征的同时,可以有效抑制噪声放大和防止出现过冲现象.

The fixed gain coefficient is used in the traditional unsharp masking algorithm(UMA), which makes it difficult to achieve a good balance between suppressing noise amplification and enhancing image features. Therefore, a two-stage unsharp masking algorithm based on the adaptive gain coefficient is proposed. Firstly, the UMA is used to enhancing the details of the image. Secondly, the weighted nuclear norm minimization(WNNM) algorithm is used to remove noise. Finally, the UMA is once again used to enhancing the strong edges of the image. In both UMAs, the gain function based on the modulus of the image gradient is used to adaptively adjust the gain coefficients at details and strong edges. Compared with the traditional UMA. the experimental results indicate the proposed method can effectively suppress noise amplification and prevent overshoot while enhancing the important structural features of the industrial X-ray image.