Variance Component Estimation in Performance Characteristics Applied to Feature Extraction Procedures

This paper proposes variance component estimation (VCE) for empirical quality evaluation in computer vision. An outline is given for the scope of VCE in the context of quality evaluation. The principle of VCE is explained and the approach is applied to results of low level feature extraction. Ground truth is only partly needed for estimating the precision, accuracy and bias of extracted points and straight lines. The results of diverse feature extraction modules are compared.

• Type: Conference
• Year: 2003
• Where: DAGM
• Authors: Marc Luxen
• Keywords: DAGM 2003, Feature Extraction, Quality Evaluation, Variance Component Estimation