From the perspective of systemic functional linguistics, this paper makes an in-depth multimodal discourse analysis of English advertising using the CNN network as a framework. By taking 100 English print advertisements randomly selected as the research object, this paper reveals the important role of image-text relationship in English advertisement. The results show that in English advertising, images and words complement each other and build meaning together. However, images in most cases are still subordinate to words, which are still the dominant means of expressing meaning. The status relationship of text and text is mainly manifested as interdependence and subordination of image to text. The logical semantic relation covers extension and projection, in which extension is mainly extended and detailed. This study not only enriches the application field of multimodal discourse analysis, verifies the practicability of relevant theories, but also provides some feasible suggestions for advertising design.