Random-Valued Impulse Denoising Using a Fast l1-Minimization-Based Image Inpainting Technique
In this chapter, an image inpainting approach based on l1-norm regularization is presented for the estimation of pixels corrupted by the random-valued impulse noise. It is a two-stage reconstruction scheme. First, a reasonably accurate random-valued impul
- PDF / 290,481 Bytes
- 9 Pages / 439.37 x 666.142 pts Page_size
- 72 Downloads / 161 Views
Abstract In this chapter, an image inpainting approach based on l1-norm regularization is presented for the estimation of pixels corrupted by the random-valued impulse noise. It is a two-stage reconstruction scheme. First, a reasonably accurate random-valued impulse detection scheme is applied to detect the corrupted pixels. Next, the corrupted pixels are treated as missing pixels and replaced by using an image inpainting technique. The inpainting method is based on the fast iterative shrinkage thresholding algorithm (FISTA). The proposed method is fast and experimental results show that it is robust to non-Gaussian and nonadditive degradations like the random-valued impulse noise. It also outperforms similar random-valued impulse denoising schemes in terms of computational complexity while preserving the image quality. Keywords Random-valued impulse noise reconstruction l1-regularization
Image inpainting Two-stage
1 Introduction Digital images are usually corrupted by impulse noise (IN) due to malfunctioning of camera sensors and distortion in communication channels during acquisition, storage, or transmission. Presence of IN in images causes degradation in image quality as well as features, like, edge details, sharpness, etc. So, removal of IN from images is very important and is an active area of research. IN is random and has two
M. Kalita B. Deka (&) Computer Vision and Image Processing Laboratory, Department of Electronics and Communication Engineering, Tezpur University, Tezpur 784028, Assam, India e-mail: [email protected] M. Kalita e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2018 A. Kalam et al. (eds.), Advances in Electronics, Communication and Computing, Lecture Notes in Electrical Engineering 443, https://doi.org/10.1007/978-981-10-4765-7_66
633
634
M. Kalita and B. Deka
common types, salt and pepper noise (SPN) and random-valued impulse noise (RVIN). Noisy pixels of an 8-bit gray-scale image corrupted by the SPN will have either 255 or 0, while the RVIN is uniformly distributed in the interval [0 255]. Removal of SPN is comparatively easier as the intensity values of the noisy pixels are quite different from the noise-free pixels. In case of the RVIN, it is often seen that the differences are not sufficient enough thereby inducing difficulty in removing it. Here, we will deal with restoration of images degraded by the RVIN. Various techniques have been proposed in this direction including the adaptive center weighted median (ACWM) [1], the directional weighted median (DWM) [2], the signal dependent rank-ordered mean (SDROM) [3], etc. In these techniques, a noise detector distinguishes noisy and clean pixels so that only the noisy pixels are restored keeping other pixels unchanged. The efficacy of such filters depends on the capability of the impulse noise detectors. As these methods fail to detect most of the noisy pixels with increase in noise levels, so the reconstruction performance is poor when the noise level is high. The problem of estimating missing informat
Data Loading...
