|
Summer School on Image Processing |
|
Project work |
|
Overview of the eye regions localization Eyes are the most prominent features on the face. The eye detection problem has been studied extensively. Reliable biometric identification of people has been an important research area for a long time. Due to the current political situation and new legislation, it has become increasingly important for example in access and border control. Currently, most of the efforts are concentrated on iris detection. The localization of the parts of the eye can also be used for the emotion recognition algorithms. Computer vision techniques have been successfully utilized in fingerprint and iris recognition, and especially fingerprint-based systems are becoming ubiquitous. Even though iris is seen as the most reliable biometric measure, it is still not in everyday use because of the complexity of the systems. One of the future challenges in the development of iris recognition systems is their incorporation into devices such as personal computers, mobile phones and embedded devices. In such applications, the computational complexity and noise tolerance of the recognition algorithm play an important role. The canthus (eye corner) is more stable than iris, since the orientation and shape of canthus are not influenced by gaze direction or eye status. However, the accurate detection of canthus is somehowmore difficult than iris detection. This is mainly because that the canthus located in skin region does not have unique grayscale characteristics like iris, and the presence of wrinkles or pouches around eyes will disturb the canthus appearances. Moreover, the canthus has a high-level, semantic definition. It is “the angle formed by the meeting of the upper and lower eyelids”. This definition makes the canthus more difficult to be fully characterized by edges, corners, or other low-level texture descriptors. |