快速位移图像高阶累积量噪点检测技术

摘要

A fast displacement image noise detection method based on higher-order cumulant is proposed.Firstly,the displacement image is fused in two-dimensional discrete wavelet transform,decomposed into different frequency domain,and filtered by using the cumulative amount of adaptive filter.The higher-order cumulant is defined in different areas of image noise,the first characteristic function of fast displacement image noise is obtained,and the natural logarithm of the characteristic function are defined as the second characteristic function,each order cumulant of Gaussian random variables for rapid displacement image noise is obtained.And then,an impulse response test model of third-order cumulants for the noisy image is established,and the fast accurate detection of noisy image displacement is realized.Simulation results show that the proposed detection method has high precision,and can effectively keep the complete image edge details.%由于快速位移图像的噪点特征与静态图像特征不同,噪点无法保留完整的图像边缘细节特征,导致传统的噪点检测方法无法对其进行准确检测,存在图像噪点检测误差大的问题.提出高阶累积量的快速位移图像噪点检测方法.先融合于二维离散小波变换将快速位移图像分解为不同的频域,利用累积量的自适应滤波器对图像进行滤波处理,在定义图像不同区域的噪点高阶累积量,获取快速位移图像噪点的第1特征函数,将特征函数的自然对数定义为第2特征函数,获取快速位移图像高斯随机变量的噪点各阶累积量,并利用上述结果组建图像噪点的三阶累积量冲激响应检测模型,实现对快速位移图像噪点的精确检测.仿真结果表明,高阶累积量的快速位移图像噪点检测方法检测精确度高,且可以较好的保留完整的图像边缘细节.