With the widespread adoption of mobile terminals in education, mobile learning has emerged as a transformative approach that enhances flexibility, efficiency, and personalization in the learning process. This study addresses key limitations of existing college English translation learning systems, including resource redundancy, outdated content, and limited adaptability, by proposing a comprehensive mobile learning and resource-sharing model based on 5G mobile communication technology. The proposed model integrates pedagogical design theory with a five-layer system architecture and employs nonlinear image diffusion algorithms to optimize content delivery. A hybrid MySQL–SQLite database system is developed to manage learner data, covering registration, login, progress tracking, and Q&A interactions. An empirical evaluation was conducted through a controlled experiment involving 64 undergraduate students, divided into experimental and control groups. Independent-sample t-test results show that the experimental group using the mobile learning system scored significantly higher (M = 91.07, SD = 4.69) than the control group (M = 84.29, SD = 9.09, p < 0.001). Furthermore, the AKAZE algorithm outperformed ORB and BRISK in image feature matching under varying conditions such as rotation, illumination, compression, and blurring, ensuring robust system performance.
