Research On Visible Light Screen Communication System Assisted By Machine Learning

With the development of science and technology,the shortage of spectrum resources is becoming increasingly serious.Therefore,the development of visible light communication has received more attention.With the continuous development of machine learning technology,it has become possible to combine machine learning with visible light communication.Screen communication is a new branch technology direction of visible light communication.In the first part of this paper,an information region positioning and tracking algorithm based on machine learning is designed.Faster R-CNN is used to locate the information region.Through training,the receiving end can be independent of the positioning image.The system calculation overhead and the LK optical flow method are used to track the information area of subsequent frames,which improves the processing efficiency of the system and is verified by experiments.This algorithm is a general pre-processing algorithm for the screen communication receiver.It is suitable for many types of screen communication systems and supports the design of new checkerboard barcodes with fewer positioning image to enhance the single-frame carrying capacity and further improve system communication.rate.Implicit screen communication aims at sensorless transmission.In the second part of this article,an implicit screen communication system is designed and implemented based on machine learning based information area positioning and tracking algorithm and QR code design.The relationship between implicit effect and embedding intensity,embedding channel,base image,grading degree and other factors were explored.Based on the analysis of human eye characteristics,the method of difference complementary frames was selected as the information hiding method,and the subjective evaluation and theoretical analysis methods were used to determine the information embedding channel as the B channel.Combined with machine learning-based information area positioning and tracking algorithm,the problems of invisible screen communication,inability to locate the information area,and possible sudden changes in the location of the hidden screen communication are solved.By designing a hybrid frame discrimination method that is suitable for implicit communication,the forward and inverse modulation frame grouping method and the embedded strength amplification method are used to reduce the system calculation overhead and improve the system stability.