Blur and Illumination Robust Face Recognition via Set-Theoretic Characterization

Abstract

blur-illumination-robust-face-recognition-computational-imaging-lab-iit-madras

Description

Results

blur-illumination-robust-face-recognition-remote-dataset-performance-computational-imaging-lab-iit-madras

Figure 12: Recognition results on the unconstrained REMOTE dataset, capturing real-world blur and illumination variation.

blur-illumination-robust-face-recognition-quantitative-computational-imaging-lab-iit-madras

Figure 5. Face recognition across Gaussian blur ➯ recognition results compared against FADEIN and LPQ approaches.

blur-illumination-robust-face-recognition-method-insight-computational-imaging-lab-iit-madras

Figure 2. The set of all images obtained by blurring a single face image forms a convex set (the convex hull of shifted copies of the image), motivating our set-theoretic approach. The paper extends this in Figure 3 to a bi-convex set once illumination variation is added alongside blur.

Publications