摘要： 结合误差分散半调噪声特征，提出一种基于偏微分方程的逆半调改进算法。通过研究偏微分方程的去噪原理，以3次B样条函数作为扩散函数，采用迭代求解偏微分方程的方法估计逆半调图像，计算每次迭代前后图像梯度模值的增量以确定平滑度的调节参数，解决偏微分方程在应用中的参数选择自适应问题。实验结果表明，该算法在图像整体平滑度和细节保持能力上都具有较好的效果。

关键词: 逆半调, 误差分散, 偏微分方程, 梯度模值

Abstract: Considering the property of error-diffused halftone noise, an inverse halftoning algorithm based on the Partial Differential Equation(PDE) is proposed. By studying denoising principle of the PDE, a normalized third order B-spline function is adopted as a diffusion one and an inverse halftoning image is obtained by solving the PDE with the iteration scheme and updating the initial image. The modulating parameter is estimated by computing the increment of the gradient magnitude between two iterated images. The problem of adaptive selection of parameters is resolved in the application. Experimental results demonstrate that the proposed algorithm has better abilities in noise smoothing and edge preserving.

Key words: inverse halftoning, error diffusion, Partial Differential Equation(PDE), gradient magnitude

中图分类号: 

• TP391