This paper proposes a new approach for the segmentation of the side face images to obtain the ear region. The proposed approach is divided into two basic steps: The first step classifies the image pixels into skin and non-skin pixels using likelihood skin detector. This likelihood image is processed by using morphological operations to detect the ear region. In the second step, image containing ear region is isolated from side face image by using one of two methods; the first is based on experiment, while the second is based measurements. The study includes a comparison of the results between the proposed study and previous ones to identify the differences. The proposed approach is applied on a database containing 146 images of 20 persons. These images were taken under different illumination, pose, day, and location variations. The partial occlusion by hair or earing was also taken in account. The results showed that the system achieved a correct segmentation with rate 95.8%.