Design And Implementatin Of Dynamic Barcode System Based On Screen-Camera Communication

With the rapid development of the hardware of electronic product and the popu-latiry of smart devices,each smart device is basically equipped with a high-resolution rear camera and screen,which provide good conditions for the study of visible light communication.The process of VLC is mainly divided into two parts:the sender and the receiver.On the sender side,the system converts the file to be transmitted into a video on the screen at a certain frame rate,then,on the receiver side,the system uses the rear camera of the smart mobile phone to capture the video on the just mentioned screen,and decodes each frame to synthesize the original file.At present,the existing VLC systems can be divided into two categories.The first is dynamic color barcode system that is visible to human eyes,its main research is to increase the throughput and improve the reliability during the transmission.The existing designs are mainly suitable for scenes that use no more than four colors that used to decode data bits,they do not give a better solution for the scenarios of more colors,such as 8 colors.For the second one,human eyes can only perceive the mixture image when the display refresh rate is beyond the critical flicker frequency that result to persistence of vision,but the cameras of smart device will produce rolling shutter effect in the same circumstance,to use the dirrerent imaging principles of the above two system,we can make the embedded pattern or barcode to be not perceived by human eyes and perceived by smart device camera.The existing systems either have to deal with the source video or the image itself,or need extra light sources,whose adjust cost is large when scene or source video switching.For the former,we design and implement a system called MMCode that solves the problem of color recognition in multiple color scenes.The system uses the designed reference color strip as the tagged data on the sender side,and use the GMM-based semi-supervised learning algorithm on the receiver side to label the tagged data and untagged data in the content area of the barcode,then the result of clustering are used to achieve color recognition.The experiment results prove that the system throughput is greatly improves by this method.For the latter,our work uses the WinAPI DirextX to implement a system called I-Code,this design makes the pattern to be added into the immediate output of the screen display,then to make the barcode invisible to human eyes,we also use the comlemen-tary color design among the front and rear frames as well as the existing system.Since our disign does not require additional eqitpment or modifications to the source file that need to be displayed,so our method is more convenient to users.