基于自适应正逆扩散算法的航海气象图降噪研究

Research on Noise Reduction of Navigation Weather Facsimile Chart Based on Self-adaptive Forward and Inverse Diffusion Algorithm

航海气象传真图通过传送后会受高斯噪声和椒盐噪声干扰而使图像变得不清晰,这不利于船舶航行安全。常用的降噪算法可明显消除高斯噪声,却无法有效消除椒盐噪声,过分降噪则又会导致传真图平滑模糊。针对这一问题,提出一个自适应正逆退化扩散模型。选用S型函数结合正逆扩散强化气象图边缘;在扩散系数中引入二阶混合偏导数,丰富对气象传真图局部信息的描述;另外添加自适应保真项以及设置自适应梯度阈值,阻止降噪过程中椒盐噪声因梯度模值较大而被强化。在Matlab中用自适应正逆退化扩散算法和常用算法对气象传真图进行降噪。对比结果表明:自适应正逆扩散算法视觉效果更好,信噪比、峰值信噪比更高,对气象传真图中的两种噪声均有很好的消除作用。

Gaussian noise and salt-and-pepper noise will be produced in the transmitting and receiving process of weather facsimile charts. It blurs the received image,which is not conducive to the safe navigation of the ship. The commonly used denoising algorithm is effective for Gaussian noise,but has no significant effect on salt-and-pepper noise. Against this problem,S-type functions are chosen to combine forward and backward diffusions so as to enhance the image edge,mixed second-order partial derivatives are introduced in the diffusion coefficient to enrich the descriptions of local information of weather facsimile chart,adaptive fidelity term is salt-and-pepper noise from being intensified due to its large gradient modulus value. In Matlab,adaptive forward and backward degenerate diffusion algorithm and traditional algorithm are used to denoise the weather facsimile chart. The contrast results show that the new algorithm has a better visual effect,higher signal to noise ratio and higher peak signal to noise ratio. It has very good elimination effect on the two kinds of noises in weather facsimile chart.