Human Ear Identification System Using Shape and structural feature based on SIFT and ANN Classifier

This paper provides an efficient methodology of human ear detection that benefits from the local characteristic of the ear and try to deal with issues due to pose, poor contrast, change in illumination, and shortage of registration. To overcome the effect of noise, and poor contrast, including illumination, it incorporates (1) image pre-processing techniques in parallel, (2) a SIFT (scale-invariant feature transform process) on images obtained to minimize the possibility of variability in pose and weak validation of images. On enhanced images, SIFT feature extraction is conducted in order to obtain local features by each enhanced image. The CCN classifier has used for the full trial to this proposed technique. The public database such as the IIT Delhi ear database, have evaluated the technique. The experimental results determined that use of the suggested fusion significantly improves the accuracy of recognition.