USB-Based High-Speed Paper-Fed Optical Scanner Upper Computer Software Research

Paper materials need to be converted into digital information in more and more industries,increasing demand for inputing and outputing.Scanner playing an important role in converting paper information into digital information,it will take much time with ordinary scanners dealing with the huge workload.So the importance of high-speed scanners is increasingly obvious.Foreign brands are in a monopoly position in the high-speed scanner market,so it is of great significance to develop a high-speed scanning system with independent intellectual property rights.This paper mainly describes the design and implementation of the software system of the upper computer of high-speed scanner.The main body of the paper contains four parts.The first part mainly introduces the working principle of high-speed scanner based on Contact Image Sensor(CIS),and the software system framework is described.The second part is about the design and implementation of the software system.TWAIN(Toolkit Without An Interesting Name)was introduced,and an image acquisition system based on TWAIN was designed and implemented.The third part mainly introduces the algorithm of image processing.A dualchannel data processing mode based on multi-thread is designed for mass data.And it introduces the designed blank page recognition algorithm,image correction algorithm,deviation correction algorithm,black edge removal algorithm,color recognition algorithm.These algorithms have been repeatedly tested with good processing effect.The blank page recognition algorithm based on contour recognition has an accuracy rate of 90% to judge whether the image is blank.The running time of test results is between 2ms 60 ms.The correction algorithm creates an independent reference table is generated for each sensor,and each pixel of the image is corrected.The correction effect is very good,the blank scan image can completely achieve all blanks.The recognition algorithm based on HSV mode has a high accuracy.It takes about 50 ms and 100 ms respectively when recognizing the A4 and A3 images.The last part tests and verifies the main functions of the software system.