Effects of severe image compression on iris segmentation performance

Abstract

The International Organization for Standardization (ISO) specifies iris biometric data to be recorded and stored in (raw) image form (ISO/IEC 19794-6), rather than in extracted templates, i.e. iris-codes. Existing literature confirms the applicability of lossy image compression in iris biometric systems, however, so far investigations on the impact of image compression on iris segmentation algorithms have remained elusive.In this work we examine the impact of severe image compression algorithms in particular, JPEG, JPEG 2000, and JPEG-XR, on the performance of different iris segmentation approaches. Experiments are carried out on an uncompressed iris database and, based on a manually annotated ground truth, effects of image compression on iris segmentation are quantified. It is found that surprisingly, JPEG causes the least segmentation errors over a wide range of high to medium bitrates (down to 0.3 bpp) despite of its weak performance in terms of PSNR rate-distortion behaviour.

Iris Image Compression Using Deep Convolutional Neural Networks

“…Bergmüller et al [ 8 ] studied the impact of using pre-compressed data in iris segmentation and evaluated the relation between iris segmentation performance and general image quality metrics. Rathgeb et al [ 9 ] investigated the impact of image compression on iris segmentation algorithms. In their work, they examined the impact of severe image compression using in particular, JPEG, JPEG 2000, and JPEG-XR algorithms on the performance of different iris segmentation approaches.…”

Effects of image compression on ear biometrics

“…Spurious compression artefacts (i.e. blocking artefacts in high contrast boundary areas) assisting in boundary localisation seem to be the reason for the performance increase of JPG [55]. It is important to note that biometric performance rates obtained in this scenario may still be practical from a forensic point of view.…”

“…These compression rates might have a beneficial effect on biometric performance. However, rate‐distortion performance may not serve as adequate predictor for biometric performance [40, 55].…”

Security Assessment of Selectively Encrypted Visual Data: Iris Recognition on Protected Samples

“…segmentation and subsequent feature extraction, which can be considered to be more realistic. It should be noted that related work has found the segmentation stage in iris recognition to be significantly affected by strong artefacts [12], such that these effects must not be neglected.…”