高速摄影下气液泡状流双阈值小波去噪方法

根据气液泡状流三维特征参数高精度测量需求,针对高速摄影法采集气液两相流图像带来的 噪声特性,提出双阈值小波去噪方法实 现气液两相流多气泡图像去噪。基 于硬阈值与软阈值的理论模型,采取双阈值小波去噪的有效方法,达到分离目标区域,保持 良好边缘特性并获得良好去噪效果 的目的,克服了传统的小波硬阈值过度“扼杀”图像信息的缺陷,优化了软阈值边缘 特性的不足,同时解决了半软阈值算法 复杂较难实现的根本问题。实验结果表明,本文方法原图像去噪后信噪比(S NR)可提高11%,熵值提高5.3%,均方根误差(RMSE) 降低了6.9%,能有效地消除图像背景噪声,在不失真的 情况下获取较为平滑的气泡图像,并保持气泡边缘特性,提高了后续流动特征的提取精度。

Double-threshold wavelet denoising of gas-liquid bubbly flow in the high-spee d photography

According to three-dimensional (3D) characteristic parameters of gas-liquid bu bbly flow and high precision measurement requirements,double threshold wavelet denoising method is proposed for such noi se characteristics to achieve gas-liquid multiple bubbles image denoising.Based on the theoretical models of hard thresho ld and soft threshold,we use double-threshold wavelet denoising method to divide target area effectively,ma intain better edge characteristics and achieve a better result.The method can solve the problems that the hard-threshold wave let denoising excessively strangles the real original information,the bubble edge characteristics cannot be optimized using soft-threshold effectively,and Semi-soft threshold algorithm is too complex to use in the practical applications.The exp eriment results show that the signal-to-noise ratio (SNR) of image after denoising can be increased by 11%,and the entropy is inc reased by 5.3%.Compared with the general wavelet threshold denoising algorithm,root mean square error (RMSE) is reduced by 6.9%.The method can eliminate background noise effect ively,smooth the images without image distortion,and maintain the edge feature of bubbles simultaneously.It impr ove s the subsequent extraction accuracy of flow characteristic effectively.