Investigation of performance trade off in High Reliability and Robust Gradient Orientation on differential sub -pixel displacement optical flow algorithms over Non-Gaussian noisy model

Abstract

This paper presents a performance analysis of 3 classical optical flow algorithms (Spatial Correlation-Based Optical Flow (SCOF), Horn-Schunk algorithm (HS) and Lucas-Kanade algorithm (LK)) under the noisy conditions. Moreover the Confidence Based Optical Flow Algorithm for High Reliability (CHR) and Robust Motion Estimation Methods Using Gradient Orientation Information (RGOI) are applied on these 3 algorithms over different characteristic of standard sequences with several Non-Gaussian noises such as Poisson Noise (PN), Salt&Pepper Noise (SPN), and Speckle Noise (SN). For HS algorithm, we also investigate the performance on the best average of smoothness weight (β) which is an important factor for the quality of outcome. These experiment results are comprehensively tested on several standard sequences such as AKIYO, COASTGUARD, CONTAINER, and FOREMAN that have different foreground and background movement characteristic in a level of 0.5 sub-pixel displacements. Each standard sequence has 6 sets of sequence included an original (no noise), PN, SPN density (d) = 0.005, SPN d = 0.025, SN variance (v) = 0.01, and SN v = 0.05 respectively which concentrated on Peak Signal to Noise Ratio (PSNR) as the performance indicator in our experiment.