Statistical analysis of pixel distribution for image distortion detection

Abstract

This study investigates the use of statistical methods to detect image distortions by analyzing pixel distributions. Using Kolmogorov-Smirnov (KS) tests and moment-based comparisons across varying sample sizes, we compare normal and distorted images to determine the minimal sample size required for reliable discrimination. Our findings demonstrate that statistical pixel analysis provides a computationally efficient alternative to traditional image quality metrics, particularly suitable for real-time applications where processing resources are limited.