The objective of the project is to verify an image in real time using only a mobile phone. Though face recognition has been implemented over the past few years on desktop computers, face verification via mobile phones (memory space and processing power) has not yet been a popular research area. Here an effort has been taken to implement face verification using a mobile phone. Using this application initially a snap shot of a face will be taken and some facts about the facial features will be saved in the mobile phone (not the image itself).Then another snapshot can be taken and compare it for verification. The ability to verify the authenticity of the user through face verification can be applied to contribute to greater security and safety for mobile devices, and the information they contain.The major objective of the project was to develop a mobile phone face verification application where it can be execute in vast range of mobile phones.
According to the functional requirements that been specified the built system must be capable of verifying face images in dynamic back grounds. This has been achieved by the system with an accuracy rate of 46%. During the secondary research its been proposed to use a mask in order to distinguish the face from the back ground. Later it was evident that this approach is highly inefficient though its effective as the face images may not be masked properly. In order to prevent that issue a virtual square has been proposed. After taking the snapshot, the user can move the square on the image using the arrow keys of the mobile phone and place it around the face region.
The application was able to fill the non functional requirements as well. One of the main non functional requirements was that the verification should not exceed 10 seconds. The application was able to verify a face image in less than a second.At the conclusion it can be stated that the proposed application has been successfully built, while full filling all the objectives and requirements specified.
As the application concentrates mainly on face verification an automatic face detection approach has not been yet implemented in the built system. As a further enhancement a face detection process can be embedded to the system, where the user may not need to fancy with the virtual square any more. During the design of the system it has been carefully assumed that such a face detection component may be useful in the future. So that in the class hierarchy the class that does the image processing has been defined to be an abstract class where a new component may extend it and further enhance the application in the future.
