Validation and reliability of the discriminative power of geometric wood log end features

Recent investigations on biometric log recognition using end face images indicated that shape information is beneficial for the biometric system performance. This study assesses the discriminative power and reliability of geometric features which are computed by means of segmented cross-sections and their pith positions. The experimental evaluation is based on cross-section images from 150 different logs, for which the ground truth of the boundary and pith position is known. By assessing the verification performance for ground truth data and automated segmentation/ pith estimation procedures this work highlights the basic discriminative power of geometric log end features and further validates their reliability in case of using automated procedures.