Noise Characterization in Web Cameras using Independent Component Analysis

An image captured by a web camera contains stationary and nonstationary noise patterns. These noise patterns are of three types i.e. Fixed Pattern Noise (FPN), Interactive Nose (IN) and Temporal Noise (TN). TN is an independent noise pattern and needs an algorithm that does exploit its higher-order dependencies. Previously, these noise patterns have been characterized using Principal Component Analysis (PCA). PCA is restricted to second order dependencies. In this paper Independent Component Analysis (ICA) has been investigated for actual TN noise. The experimental results demonstrates the effectiveness of the proposed method.