压缩感知下梯度投影在图像去噪中的应用

图像传输和接收的过程中不可避免地受到多种因素的影响而产生噪声，这样通常使得图像模糊不清难以辨识。针对去除图像高斯白噪声对图像辨识的干扰，提出利用压缩感知最优化求解思想，运用一种快速计算方法将高维数的图像信号投影在低维空间上，从而类比于压缩感知理论进行重构处理，并实现对含噪声信号的图像进行还原。实验结果证明，梯度投影法能较好地去除图像高斯白噪声，尤其对于图像中物体边界的辨识效果明显，为图像去除噪声提供了新的思路和方法。

Gradient Projection Algorithms for Image De- Noising Based on Compressed Sensing

Image transmission and reception process is inevitably affected by many factors and noise, which typically makes the image blurry illegible. For removal of image Gaussian white noise disturbance on the image recognition, this paper proposes the use of compressed sensing optimization for solving ideas that a quick calculation method using the high- dimensional projection image signal in a low- dimensional space, thus analogous to the theory of compressed sensing reconstruction process and to achieve an image signal containing noise reduction. Experimental results show that the gradient projection method can be better to remove the image Gaussian white noise, especially for the identification of objects in an image boundary. For the image de- noising provides new ideas and methods.