The Palmprint Recognition System Based On Android Platform

Android has leapt capita ranked first with its free development tools and opensource platform for mobile devices. As security protecting program of mobile devicewith android system, the accuracy of face recognition is low and the robustness is weak.There is a big problem of security risks in the research of fingerprint identificationbecause of its contactless, besides, the location of the sensor restrict user to use.Palmprint identification is widely used in security protection, access control andattendance system due to its accuracy is high and the cost of its device is usually low,besides, the feedback from the users is much better than other technology. As manyknows that most of the virus spread by the contact, and the security risk caused bycontact is the biggest drawbacks. Hence, we choose the contactless palmprintidentification and the main content is listed as follows:Build the environment run system, design and implement frame of this project.Build the android platform as an operating system; programming JNI to offer a interfaceto call C library; design and implement the frame of software development, adopt threelevel which is application development, UI development and library call development.It is a layout of the system, contain preview, processing, sampling and stop preview, andas an interface to be called by application development. As a response of the processmodule, library call development is a kernel of algorithm in this project.Improved threshold processing. From the learning of threshold processing basedskin color used in face recognition and OTSU based skin color algorithm, we raise anew improved OTSU based skin color algorithm. The algorithm raise in this articleadapt all the devices based android system which is own fill light. And it can segmentthe object from the black and light background. Besides, it can solve the illuminationsensitive issues and the background which color is similar to the skin.Algorithm validation and results analysis. Design some experiment to test thesegmentation in gray picture and color picture, in fix threshold and in self-adaptedthreshold, based color similarity and based color self-adapted threshold. And verify thealgorithm which is raised in this article from two hands, one is the accuracy ofsegmentation, and other is the consuming time of excution. And at last, verify theimportance of the preprocessing periods by the accuracy of identification.