| At present,the main implementation of automatic testing platforms for mobile terminal style testing is to build a virtual mobile environment,perform simulated clicks,generate testing images,and perform image screening by testers to find testing results that do not meet expectations.This automatic test combined with manual regression is difficult to truly save testing manpower.In order to solve the problem of semi-automatic and semi-manual testing,the automatic testing platform designed in this paper incorporates a pattern verification module,which does not require testers to perform full image screening to save testing manpower.This platform uses Python language to implement the overall framework.The platform adopts the concept of componentization,including three levels: environmental operation layer,logic processing layer,and application program layer.It implements the decoupling of tools and business logic to ensure low coupling of the platform and improve work performance.Analyzing the needs of testers to achieve automatic test of mobile advertising styles and interactive functions,design and implement testing task triggering module,configuration updated module,style rendering module,style verified module,and online revenue monitoring module.The five modules work together to analyze the configuration information,testing case information and other information of the testing environment module input by the user,create a new testing task,schedule mobile phone resources for style displaying,style capturing,and template matching of the generated style images to determine similarity degree.Among them,when scheduling resources,in order to ensure compatibility with screens of different sizes,RR Algorithm is used to select four mobile phones of different brands for style capturing.When template matching with the reference picture,choose Hu invariant moments to calculate eigenvalues and Hausdorff distance algorithm to calculate the similarity,to improve the success rate of automatic testing result comparison.In order to realize this automatic testing platform,OCR Recognition Technology,Hu invariant moments and Hausdorff distance Algorithm are all deeply studied.The requirement analysis,design,implementation and test of the test platform are also carried out to complete the development of the test platform.By comparing the new version of automated testing platform which combines Hu invariant moments with Hausdorff distance Algorithm for pattern verification with the old version of automated testing platform which use the semantic cutting for pattern verification,the conclusion can be drawn: In this paper,the automatic test platform combined Hu invariant moments with Hausdorff distance algorithm for template verification has improved the success rate of automatic comparison testing results in the testing requirements that support the rapid iteration of mature software that has been developed,and the stability and efficiency of test have been increased from 40% to more than 90%,which is more suitable for practical engineering applications. |
PDF Full Text Request