The Design And Implementation Of AD Templat Automated Testing Tool Based On Image Matching

Ad templates test is produced by a commercial branch in the Internet industry.Its implementation of automated testing is highly anticipated by testers. The thesisdiscussed is the design and implementation of testing tools that can test ad templatesautomated. It comes from the requirements of E-commerce search sector testingteam in Baidu. The ad templates automated testing tools can reduce test costs andimprove test efficiency.Throughout the project implementation process, designed and developedautomated testing tools which meet the the needs of ad templates testing foradvertising template test features. This paper describes development process of thead templates automated testing tool from the tool’s requirements, design,implementation, and testing.The thesis discussed the features of existing web automated testing tools firstly.The ad templates test automation requirement and overall architecture of the adtemplates automated testing tools are extracted after the summary of ad templatestest-related business and the testing process.The design and implementation of tool are accomplished based on the overallarchitecture. The ad templates test tool’s task scheduling algorithm came into beingfrom existing scheduling algorithms in task management module. The thesis alsogives a scheduling algorithm evaluation strategy. Based on API provided with theselenium framework, adopt the way of injecting JavaScript code to the browsers toimplements the ad HTML page show, HTML page parsing and advertisementscreenshots modules. Image matching is advertisement screenshots and referenceimage matching. Image matching based on the theory of genetic algorithm. Testresults of the task execution come from the results of image matching. In order toimprove the accuracy of image matching, Bilateral Filter algorithm is used toeliminate image noise before the matching.Finally, Ad templates automated testing tool accomplished and tested under thecoordination of the various functional modules. The completed Ad templatesautomated testing tool can be used normally and reached a preset target. Test taskscheduling algorithm achieves the expected rate of the task processing; at the same time ensure the test task processing clients’ CPU utilization is relative to the average.Image matching algorithm can present advertisement screenshots and referenceimage match results correctly and effectively. The test task execution results andexpected results are consistent.