A novel automatic ear anatomy feature extraction method was proposed.Firstly,the side face image including an ear was preprocessed to eliminate the noise and spur.Secondly,we regarded the 2D gray image as a surface because the characteristics of surface are that gray level changes in edge area is sharp and the curvature is larger than that of the smooth area.In accordance with the characteristics,the maximum and minimum principal curvatures of surface were calculated,and then,automatic threshold was used to segment curvature images.Finally,two thresholded images were combined to get the feature edge image.The experiments demonstrate that extracted feature edge curve is smooth and connected.Experiment shows that our method is robust to illumination intensity and edge curve can meet the requirements of next process.