The rapid development of social networks has promoted the exchange of a large amounts of data, but also the spread of false information. Many research works have addressed the detection of rumors, mostly by analyzing the textual content of messages. However, the visual content, especially images, remains ignored or little exploited in the literature. Yet, visual data are very popular on social media and their exploitation proves important for analyzing rumors. In this paper, we present a synthesis of the state of the art about rumor classification and summarize the main tasks of this process, as well and the approaches to analyze this phenomenon. We particularly focus on the techniques adopted to verify the veracity of images. We also discuss the datasets used for rumor analysis and present the research leads we plan to investigate.