Research On Digital Image Aspect Ratio Correction Technology

With the rapid development of science and technology,information technology represented by communication technology,computer technology and network technology has increasingly penetrated into every aspect of social life.The integration of information technology and the process of education and teaching gives birth to some new teaching methods,teaching models and teaching theories.As an important form of information presentation,digital image has been applied to teaching resources such as teaching courseware and teaching web pages.At the same time,it has also become an important form of knowledge expression in various academic papers and textbooks.Digital images not only simplify the explanation and description of complex knowledge,but also highlight the key points in the teaching content,enabling learners to obtain a large amount of useful information by converting text memory into image memory,and also deepen their understanding of the knowledge that have learned.However,it often finds some images with serious aspect ratio imbalance in teaching courseware,teaching web pages and academic reports.Some of these images with distorted aspect ratio are caused by the improper operation of the producer or editor,and some are caused by the original image.The emergence of such problems affects the normal presentation of teaching content,it is not conducive to the accurate transmission of teaching information,and will also have an impact on the teaching effect,sometimes,it makes learners misunderstanding the content presented by images.Therefore,it is necessary to correct the aspect ratio of digital image effectively so that it can express information normally.Aiming at the problem of "how to automatically optimize and correct the aspect ratio of digital image",in this paper,we propose two kinds of digital image aspect ratio correction techniques.Firstly,we propose a digital image aspect ratio correction technique based on Img RV.By extracting edge features from digital images and drawing density map,the relationship between edge points can be expressed as a function.Then,the aspect ratio is selected from the density map to represent the digital image again.Finally,the aspect ratio selection method for curves and scatter graphs is successfully extended to digital images to realize aspect ratio correction of digital images.Secondly,we propose a digital image aspect ratio correction technique based on Goog LeNet.The technology will be introduced the convolutional neural network Goog LeNet to the digital image aspect ratio correction process,by using a set of image ready to Goog LeNet training,study the inherent law of digital image aspect ratio correction from a large number of training data,and analysis convolution layer network training process to image characteristics from the Angle of characteristic graph visualization.The experiment shows that the proposed method can effectively solve the problem of digital image aspect ratio correction by extracting the edge features,color information,texture features and more abstract features of the image,learning the law of aspect ratio correction from many images.Digital image aspect ratio correction technology can not only help educators to use images to express accurate teaching knowledge,provide visual and intuitive teaching resources for teaching,but also enable learners to obtain better visual experience and reduce the cognitive load of reading images.