Abstract
In this paper, we investigate the denoising of image sequences i.e. video, corrupted with Gaussian noise and Impulse noise. In relation to single image denoising techniques, denoising of sequences aims to utilize the temporal dimension. This approach gives faster algorithms and better output quality. This paper focuses on the removal of different types of noise introduced in image sequences while transferring through network systems and video acquisition. The approach introduced consists of motion estimation, motion compensation, and filtering of image sequences. Most of the estimation approaches proposed deal mainly with monochrome video. The most usual way to apply them in color image sequences is to process each color channel separately. In this paper, we also propose a simple, accompanying method to extract the moving objects. Our experimental results on synthetic and natural images verify our arguments. The proposed algorithm’s performance is experimentally compared with a previous method, demonstrating comparable results.
Highlights
Denoising of image sequences is one of the critical tasks of image processing and signal processing
Fourier transformed based approaches [2] are used to deal with color image sequences for motion estimation
For the addition of is determined by the ωI (x) function into the bilateral filter, we have to determine the strength of radiometric component in the impulse noise
Summary
Denoising of image sequences is one of the critical tasks of image processing and signal processing. Denoising image sequences extends the above operations to handle the sequential dimension. Algorithms for the denoising of image sequences aim to remove such type of noise while utilizing both the spatial and temporal domains. A suggested approach that utilizes the temporal redundancy is motion compensation [2, 7,8,9,10]. Most of the motion estimation approaches deal only with monochrome images. Fourier transformed based approaches [2] are used to deal with color image sequences for motion estimation. The proposed approach further computes motion parameters and filter weights by maximizing peak signal to noise ratio. Different types of video sequences are used to test the filter performance in terms of visual quality and peak-signal-to-noise ratio (PSNR)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
