广义S变换域低照度图像时频分析及滤波

摘要

为降低二维广义S变换的计算复杂度和内存占用度,基于二维广义S变换时频分析原理,提出了一种改进快速离散正交S变换算法(FDOST).采用FDOST算法对低照度图像进行S变换,分析了高斯噪声和原始图像在广义S变换域的时频分布,给出了区分原始图像和高斯噪声的方法以及基于改进广义S变换低照度图像时频滤波法.对合成含噪图像和实际低照度图像进行去噪仿真,结果表明:基于改进广义S变换FDOST滤波方法可去除图像中的高斯噪声,去噪后图像信噪比较去噪前提升了6%,最大化保留了原始图像信息.%In order to reduce computational complexity and memory occupancy of two-dimensional generalized S transform,an improved fast discrete orthogonal S transform (FDOST ) algorithm is proposed based on time-frequency analysis of two dimensional generalized S transform.Low illumination image is transformed with S transform using FDOST algorithm.Time-frequency distribution of Gauss noise and original image are analyzed in generalized S transform domain.The method of distinguishing the original image and Gauss noise and low illumination image time frequency filtering method based on the improved generalized S transform is presented.The denoising of synthesis of the noisy image and the actual low illumination image is simulated.The results show:FDOST filtering method based on the improved generalized S transform removes the Gauss noise in the image;After denoising,SNR of the image is improved by 6%,while the retention of the original image information is maximized.