Digital image processing (DIP) has a much broader scope of tools, techniques and algorithms than the corresponding analog image processing. This valuable advantage can be invested in processing and restoring informative image out of distorted one, for instance noisy image. The main objective of this paper is to insure accurate quality metrics to measure spatial filtering techniques in estimating an actual image out of noisy data. The image is distorted with Salt & Pepper noise. The proposed restoration techniques include Mean filter, Median filter, Max filter, Min filter and Wiener filter. The filters are investigated versus the grayscale image. Different quality metrics are employed, including, Mean Square Error (MSE), Performance Index (PI), Peak Signal-to-Noise Ratio (PSNR), and Image Enhancement Factor (IEF). Moreover, blind metrics that include BRISQUE and NIQE are studied and their accuracies get assessed. Constantly, the investigated quality metrics show that Median filter demonstrates a greater outcome than all of its counterparts. Study shows also that the accuracies of quality measures are in agreement with each other as well as with the subjective assessment of the restored images.
Abdussalam Alhadi Addeeb, Ahmed Abdallah Aburas (2019); DSP Filters and Quality Metrics for Image De-noising; International Journal of Scientific and Research Publications (IJSRP)
9(3) (ISSN: 2250-3153), DOI: http://dx.doi.org/10.29322/IJSRP.9.03.2019.p8755