Image Quality Assessment in Forensic Facial Comparison

n forensic science, the reliability of facial comparison heavily depends on the quality of the images analyzed. This thesis examines the role of Image Quality Assessment (IQA) in facial comparison, focusing specifically on how different im- age characteristics impact the accuracy. Existing general-purpose IQA methods are evaluated, revealing significant limitations when applied to realistic forensic scenar- ios such as surveillance footage or uncontrolled imaging environments. To address these issues, a specialized IQA framework tailored explicitly for facial comparison tasks is proposed, integrating objective computational metrics and subjective as- sessments from trained examiners. The experimental evaluation conducted and demonstrates that the proposed framework is strongly correlated (up to 83%) with examiner decisions, outperforming standard IQA metrics by approximately 25%. These findings underline the importance of developing standardized IQA protocols tailored for forensic workflows, thus enhancing the reliability of facial comparisons and supporting higher evidentiary standards in identity verification contexts.