Higher quality of infrared images for future network security systems

It is well known that infrared (IR) imaging has been used extensively for various applications, in particularly in those dark conditions. Therefore, it has been widely employed in various tracking systems, in particularly, with the Internet of Things (IoT), sensors integrated with cloud computing, which makes IR even more popular for future networks. But various images, including IR images almost cannot avoid contaminations from various noises sources even when they were born. High quality images are particularly needed in those applications, especially for the security and military environments. In order to obtain a good quality of IR image for the future network security, in our current paper, it proposes a novel denoising algorithm based on Cohen-Daubechies-Feauveau wavelets followed by the low-pass filters with the length 9 and 7 (CDF 9/7) wavelet transform. In our algorithm, we first applied the lifting structure to improve the drawbacks of a traditional wavelet transform. As the normal wavelet transform appears to be restricted and limited for a class of opportunities of multi-scale representation of multi-dimensional signals to estimate the noise level and then remove them from original images, which is less efficient and effective. In contrast, our proposed algorithm in this paper is much efficient in estimating and reducing noises from the images contaminated by multi noises, such as Gaussian noise, Poisson noise, and impulse (Salt& pepper) noise. Experimental results on several tests for infrared images by using our algorithm are presented, as an example, under the noise with standard deviation σ = 0.2 and density = 20% cases, for peak signal to noise ratio (PSNR) increasing 14%; mean square error (MSE) our method decreasing 83%; and mean of structural similarity (MSSIM) increasing 67% under the same conditions. Obviously, the results strongly support the fact that our proposed algorithm is significantly superior to other related methods.