Research On GUI Layout Evaluation Method Based On Intelligent Object Recognition

With the development of computer,communication and other technologies,graphical user interfaces(GUIs)have been widely used in various industries and people's lives.What's more,GUIs can help people improve work efficiency and make life better.The design and evaluation of GUIs have also been continuously valued.The design elements of the graphical user interface include color,content,structure,layout and so on.The layout of the elements is particularly important during the evaluation process.At the same time,the rapid development of information technology and artificial intelligence technology has put forward higher requirements for GUI evaluation,and also provided more potential application methods.Therefore,the interface layout evaluation is taken as the research subject in this study,and quantitative evaluation methods of interface layout are constructed.At the same time,combined with artificial intelligence technology,the automatic quantitative labeling of indicators based on intelligent identification of objects has been achieved.Finally,comprehensive evaluation models of interface layout are constructed and verified.The main work of this thesis are as follows:First of all,through reading,sorting and summarizing the literature related to interface layout evaluation,the interface layout influencing factors were determined and a quantitative evaluation method for the interface was constructed.By comparing the evaluation method based on actual subjects with the evaluation method based on statistical models,the new requirements of interface evaluation in the intelligent era are analyzed.At the same time,the application of computer vision technology in the field of graphical user interface is summarized.Secondly,segment the graphical user interface was segmented intelligently and the design elements in the interface were extracted based on the technology of computer vision.And the edge information of its circumscribed rectangle is obtained,including the size and coordinate information of the rectangle.By substituting the obtained edge information data into the quantitative formula of interface layout evaluation,the automatic calculation of interface layout indicators is realized.Through the analysis of the business requirements,functional requirements and interface requirements of the interface evaluation software,the prototype of the interface layout automation evaluation system is designed.Finally,the construction and verification of the comprehensive evaluation models based on the layout metrics are carried out.Taking homepages of news webpages as examples,a weighting method based on entropy is proposed and a comprehensive evaluation model of interface layout.Meanwhile,the user's subjective evaluation of the interface layout design through questionnaire has been obtained.Then it uses multiple regression to fit the relationship between the objective metrics and the user's score to build a comprehensive evaluation model.Finally,the ranking method in the field of psychophysics is used to verify the effectiveness of the two methods.All in all,the time,labor and material costs of GUI interface evaluation can be effectively reduced when the method proposed in this thesis is used.Furthermore,the usability,compatibility and flexibility of interface layout evaluation can be improved as well.The results of this study can be used in the actual development process.This evaluation can be placed in any part in the interface design.The screenshot can be used to evaluate the quality of the interface layout,improve the design,and accelerate the development progress.In the future,this method can introduce more metrics and modules,and expand functions as needed.