Performance Analysis of Eye localization Methods for Real Time Vision Interface using Low Grade Video Camera

It is necessary to correctly and precisely achieve eye localization, which is a fundamental step for the initialization for other eye localization based applications. There are various methods including special equipment based methods and image based methods to perform this task. Special equipment based methods are very accurate but not practical for day to day use. Image based approaches are user friendly, allows free head movement, avoids specialized hardware and infrared exposure but more difficult to implement. Performance is analysed for state of the art eye localization methods for real time vision interface using low grade camera that use similar objective criterion for error measurement on standard dataset for fair judgment. Finally their localization results are compared based on various error values and rank.