Owning to the rumpled nature of the human ear, creating wonderful hills and valleys biometric authors have shifted to ear biometric in the quest of establishing it to the likes of fingerprint, iris, and face based biometric system. This paper avail a little summary view of ear biometric till date and proposes the use of gradient histogram features for recognizing the ear. The ridges-like structure of human ear provides vital information which can aid for its recognition with or without any much enhancement processes. After the image is comfortably cropped to reduce background noise from the hair, the image is further enhancement to remove some spike of noise that could cause the failure of the proposed method since allowing undesired information going into for processing will be erroneously expensive. Furthermore, the gradient of the enhanced image is computed and subsequently its Gradient Histogram (GH) is built. A dimensional reduction technique Principal Component Analysis (PCA) is used to reduce the feature vector space presented by the gradient histogram. Using Multilayer Feed Forward (MLFF) Neural Network classifier, the experimental result showed 100% recognition accuracy on USTB ear database.