As more mobile applications become available to a broader user base, the need to localize applications to various languages and cultures grows. The need for localization demands localization testing. In this paper, the user interface localization problems are identified and categorized. The automated testing method containing automated localization defects detection algorithms using applications’ screenshots analysis is presented. The proposed method was validated experimentally by comparing automated testing and manual review. The experiment contained an automated and manual review of 781 Android applications. The manual review results were compared with the proposed automated testing method results. The comparison shows that automated methods can save localization testing time and discover more defects. However, the proposed method does not fully substitute manual testing but can act as a helper.

Mobile and web applications must operate and be displayed correctly on many different devices and browsers. The visual testing of web or mobile applications is usually a manual process that requires a significant amount of testing time, meaning that applications are tested only on a few devices. It is then assumed that the applications will be displayed correctly on other compatible or similar devices. This paper presents an automated visual testing method for user interfaces. The main contributions of this paper are a classification scheme for visual defects of user interfaces and the definition of an automatic visual testing method that tests applications on many different devices with varying hardware and software parameters. The method is based on an automated search for defects using heuristic and expected state prediction algorithms, which involves analyzing the resources used by applications and screenshots. The testing method works by executing applications on a full set of devices, taking a screenshot at every execution step, and analyzing each of these screenshots. The manual as well as automated testing approaches were validated on 781 of Android applications. The experimental results show that the proposed method has advantages over manual testing.