In the context of emerging engineering education, physics competition teaching should not only serve the function of talent selection but also emphasize the cultivation of students’ modeling ability and computational thinking. When dealing with increasingly complex multi-element optical imaging systems in competitions, traditional ray-tracing methods often involve cumbersome calculations, which may obscure the underlying physical picture. This paper introduces matrix optics as a bridge between physical modeling and mathematical algorithms. By comparing the logical paths of traditional geometrical methods and matrix-based methods, and by taking the first problem of the 40th National High School Physics Competition Semifinal as an example, this paper demonstrates the advantages of the matrix method in analyzing “black-box” optical systems. Furthermore, the paper explores how Python can be used to assist matrix operations and realize a teaching progression from manual derivation to computer simulation. This study provides a new perspective and practical approach for integrating computational thinking into physics competition teaching.