Software Development Of Real-Time Precise Printing System Based On Embedded GPU

The traditional digital printing directly prints the pattern on the fabric.Such printing mode is simple,but cannot print accurately to match the distortion of non-planar fabrics like embroidery and lace.Furthermore,for fabrics that require double-sided printing,such printing mode cannot print accurately on the opposite side of the same spot on the fabric.Therefore,the development of precise printing systems for non-planar fabric printing and double-sided printing has high engineering application value.Hence this thesis has developed a real-time precise printing system software based on embedded GPU.This system software is divided into five parts:image acquisition module,image correction algorithm module,communication module,printing module,and log module.The image acquisition module collects camera images and receives standard images from PC.The image correction algorithm module performs corner detection on the acquired image and standard image based on the Shi&Tomasi algorithm and sub-pixel corner points,and performs brute-force matching based on ORB feature descriptors.After selection by the K-Means and RANSAC algorithms,it uses the TPS algorithm to correct the standard image.The communication module enables the intercommunication of different hardware through TCP/IP,PCI-E,SPI and GPIO.The printing module sends print data to the FPGA after interpolation,and controls the nozzle through the FPGA.The log module records the operating status and abnormalities of the system,and displays them through the client and a small web server.Tests show that the real-time precise printing system software based on embedded GPU developed in this thesis can accurately perform non-planar printing and double-sided printing,which meets the system's functional requirements.The average correction accuracy of the image correction algorithm has reached 98.22%and the average RMSE is 1.52 pixels.It takes an average of 1.465s to process 1 PASS image,which meets the system's performance requirements.