Assessment on Image Quality Changes as a Result of Implementing Median Filtering, Wiener Filtering, Histogram Equalization, and Hybrid Methods on Noisy Images

The information content in an image can experience a decrease in quality due to noises. Accordingly, noise removal and histogram equalization (HE) are among the processes used to enhance image quality. The purpose of this research is to determine the effect of changes in image quality as a result of applying median filtering (MF), Wiener filtering (WF), HE, or hybrid methods on noisy images. Here, face images are used as input. The research stages began with preprocessing, i.e., cropping, size normalization, and color-to-gray conversion. Then, Gaussian and salt-and-pepper noises with intensities of 20%, 50%, and 80% are applied to the images. Consequently, the processes, i.e., HE, MF, WF, hybrid method 1 (HE and MF), and hybrid method 2 (HE and WF), are implemented, and then the image quality is evaluated. The results of the research show that the best methods to enhance Gaussian and salt-and-pepper noisy images are WF and MF, respectively. HE and the two hybrid methods generally do not improve image quality. The implementation of hybrid method 2 results in the maximum structural similarity index with 80% Gaussian noise. Meanwhile, MF provides the minimum mean-square error and maximum peak signal-to-noise ratio with 20% Gaussian noise.