As software systems expand in complexity, managing the vast and varied collection of test cases becomes increasingly difficult with traditional manual testing methods. This paper presents a new approach for automating the generation of structured test cases, named Test Element Extraction and Restructuring (TEER), which leverages the advanced natural language processing capabilities of large language models (LLMs). Specifically targeting human-computer interaction (HCI) software, TEER employs prompt tuning techniques to extract critical elements from natural language test cases and systematically reassemble them into structured formats. The study evaluates the effectiveness of TEER by applying it to common test cases from desktop HCI applications. The experimental results demonstrate that this method successfully produces structured test cases that meet predefined requirements.