Ear Photo Recognition using Scale Invariant Keypoints

K. Dewi and T. Yahagi (Japan)


ear photo recognition, scale invariant keypoints, force field transformation, PCA.


Various approaches, such as Principal Component Analy sis (PCA) and force field transformation, have already been researched for ear photo recognition. Force field transfor mation is invariable to scale and rotation. However, the number of extracted features is considerably insufficient for recognition. In order to cope with this problem, we pro posed ear photo recognitionusing scale invariant keypoints. The keypoints are extracted by performing Scale Invariant Feature Transform (SIFT). In our experiments, SIFT gener ates approximately 16 keypoints for each ear image. After we extract the keypoints, we classify the owner of an ear by calculating the number of keypoint matches and the av erage distance of the closest square distance. We compared our results with ear photo recognition using PCA and ear photo recognition using force field feature extraction. Our experimental results show that ear recognition using SIFT gives the best recognition result.

Important Links:

Go Back