Multiscale-SSIM index based stereoscopic image quality assessment

Abstract

Stereoscopic image quality typically depends on two factors: i) the quality of the luminance image perception, and ii) the quality of depth perception. The effect of distortion on luminance perception and depth perception is usually different, even though depth is estimated from luminance images.Therefore, we present a full reference stereoscopic image quality assessment (FRSIQA) algorithm that rates stereoscopic images in proportion to the quality of individual luminance image perception and the quality of depth perception. The luminance and depth quality is obtained by applying the robust Multiscale-SSIM (MS-SSIM) index on both luminance and disparity maps respectively. We propose a novel multi-scale approach for combining the luminance and depth scores from the left and right images into a single quality score per stereo image. We also explained that a small amount of distortion does not significantly affect depth perception. Further, heavy distortion in stereo pairs will result in significant loss of depth perception. Our algorithm performs competitively over standard databases and is called the 3D-MS-SSIM index. Index TermsStereoscopic images, full-reference image quality assessment, depth perception, MS-SSIM.

Memristive system design for variable pixel G-neighbor denoising filter

“…The improved metric to simultaneously compare noise reduction and structural similarity preservation is SSIM [34]. It combines comparison of luminance (l), contrast (c) and structure (s).…”

Quality assessment metrics for edge detection and edge-aware filtering: A tutorial review

“…SSIM is more closely related to the human visual system as it extracts useful information as luminance (l), contrast (c) and structure (s). It can be applied to evaluate structure preservation and noise removal [13].…”