基于贝叶斯阈值优化选取的图像动态降噪算法  

Dynamic Image Denoising Algorithm Based on Bayesian Threshold Optimization

作　　者：於平[1] YU Ping(Chuzhou City Vocational College,Chuzhou Anhui 239000,China)

机构地区：[1]滁州城市职业学院,安徽滁州239000

出　　处：《佳木斯大学学报（自然科学版）》2023年第4期60-62,94,共4页Journal of Jiamusi University：Natural Science Edition

基　　金：2021年滁州城市职业学院青年教师项目(2021qnxm08);2023年滁州城市职业学院科研重点项目(2023Skzd03)。

摘　　要：小波阈值去噪通常采用阈值法,导致其图像降噪效果不佳、失真效果未明显改善。因此,基于贝叶斯阈值优化选取策略确定小波阈值,实现高精度图像动态降噪。贝叶斯估值算法基于概率学统计方法自适应选择噪声阈值,采用QPSO算法确定贝叶斯阈值的可调参数;QPSO算法借助量子行为原理引入“粒子最优位置”变量,使用动态惯性权重与学习因子策略,优化粒子群全局搜索可调参数的能力。实验结果显示,该方法降噪后的图像噪声点明显减少、失真问题得到改善、叶片边缘清晰;同时降噪后图像获得了更高的信噪比,在保留图像细节、抑制噪声方面展现了良好优势。Wavelet threshold denoising usually adopts threshold method,which leads to poor image denoising effect and no obvious improvement in distortion effect.Therefore,wavelet threshold is determined based on Bayesian threshold optimization selection strategy to achieve high-precision image dynamic denoising.Bayesian estimation algorithm adaptively selects noise threshold based on probability statistics method,and uses QPSO algorithm to determine the adjustable parameters of Bayesian threshold.QPSO algorithm uses quantum behavior principle to introduce"particle optimal position"variable,and uses dynamic inertia weight and learning factor strategy to optimize the ability of particle swarm to globally search for adjustable parameters.The experimental results show that the noise points of the image reduced by this method are significantly reduced,the distortion problem is improved,and the blade edge is clear.At the same time,the denoised image has a higher signal-to-noise ratio,which shows good advantages in retaining image details and suppressing noise.

关 键 词：贝叶斯阈值 小波 QPSO算法 粒子 量子 降噪 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]