Robust Degraded Face Recognition Using Enhanced Local Frequency Descriptor and Multi-scale Competition

TL;DR

This paper introduces an enhanced local frequency descriptor and a multi-scale competition strategy to improve degraded face recognition, especially in low resolution and blurred images, demonstrating promising results on Yale and FERET datasets.

Contribution

It proposes a novel Enhanced LFD that jointly utilizes spatial and frequency information, and a multi-scale competition strategy for robust face recognition.

Findings

Improved recognition accuracy on Yale and FERET datasets.

Enhanced robustness to low resolution and blurred images.

Effective multi-scale feature extraction strategy.

Abstract

Recognizing degraded faces from low resolution and blurred images are common yet challenging task. Local Frequency Descriptor (LFD) has been proved to be effective for this task yet it is extracted from a spatial neighborhood of a pixel of a frequency plane independently regardless of correlations between frequencies. In addition, it uses a fixed window size named single scale of short-term Frequency transform (STFT). To explore the frequency correlations and preserve low resolution and blur insensitive simultaneously, we propose Enhanced LFD in which information in space and frequency is jointly utilized so as to be more descriptive and discriminative than LFD. The multi-scale competition strategy that extracts multiple descriptors corresponding to multiple window sizes of STFT and take one corresponding to maximum confidence as the final recognition result. The experiments conducted on Yale and FERET databases demonstrate that promising results have been achieved by the proposed Enhanced LFD and multi-scale competition strategy.