Research And Design Of Network Scoring System Based On Image Recognition Technology

The rapid start of Internet application technology has made online education get an unprecedented development.At present,the main development direction of the online education is based on the incorporation of actual university and the virtual campus.Internet education has a huge market in developed countries,which indicates that the network marking industry is also developing rapidly.Both the automatic paper scanning identification system and the fast marking on the network platform can speed up the application of the network in learning,and thus become the driving force for the informatization and modernization of the entire network education.In this dissertation,the writer firstly does pretreatment operations of grayscale processing of the character images,binarization of the target area image,discrete noise and smoothing,geometric correction and image segmentation,the normalization of characters.Secondly,the writer mainly focuses in the research that based on application of neural network,classifier's design into scanned handwritten.What's more,the writer compares traditional classifier to neural network classifier.Finally,the results of the experment indicates that neural network classifier can recognise the scanned handwritten letters ABCD very well.The network marking system is designed by VC ++ programming.Based on the huge number of experimental samples,the final correct recognition rate of the system is up to 97.0% with an average of 93.92%.