Research On Mobile Phone Recognition Software And Key Algorithms Of Halftone Dot Image

Using halftone screening technology to realize information hiding has the advantages of low cost,good effect and high reliability.It is widely used in the field of printing anti-counterfeiting.The halftone secure QR code formed by embedding the halftone pseudo-random noise with hidden information into the ordinary QR code has good anti duplication performance while realizing information hiding.In order to improve the reading efficiency and user experience of this halftone secure QR code,this thesis studies the halftone dot mobile phone recognition software and key algorithms.Relevant research work is as follows:(1)In order to solve the problem of insufficient image sharpness in the process of halftone QR code image scanning,the commonly used image sharpness evaluation algorithms are studied and compared.According to the characteristics of regionalization of microstructure characteristics of halftone image and obvious contrast between adjacent pixels,the improved algorithm based on the least square method is adopted to calculate the image gray gradient to evaluate the sharpness of collected images.According to the characteristics of large edge gray gradient change of QR code image,combined with Pulse Width Modulation(PWM)waveform,the spectrum of QR code image is analyzed,and a single parameter sharpness evaluation model based on highfrequency ratio is established.Experiments show that the two improved algorithms can quickly screen clear halftone QR code images.(2)In order to meet the high requirements of detection accuracy,response speed and storage space in the application of halftone secure QR code,the mainstream target detection algorithms at this stage are studied and analyzed.Yolov5 n with small volume is selected as the basic model for improvement.The Stemblock structure is used to replace the traditional convolution for down sampling,which not only ensures the strong feature expression ability,but also reduces the amount of parameters;The basic structure of Shuffle Netv2 is used for feature extraction to reduce group convolution and point-to-point addition,while maintaining the consistency of the number of input and output feature channels of convolution layer,so as to improve the running speed of the model.The experimental results show that the improved lightweight network can better meet the needs of halftone QR code detection.(3)In order to meet the needs of further promotion and application of halftone secure QR code,the mobile phone recognition software of QR code is studied.Based on the secondary development of Zxing open source package of Google,the Android Native Development Kits(NDK)is used to process and recognize halftone images,and the optimal design of mobile phone recognition software of halftone secure QR code is completed.The identification result of halftone security information can be output while reading ordinary QR code information.