With the recent advances in technology, we have the opportunity to study the city dynamics at a large scale. A fundamental step is to be able to sense extensive areas. Participatory sensor networks have the potential to become a very fundamental tool to study social behavior at a large scale. Currently, there are some location sharing services, such as Foursquare, that record the user's location along the time, which is clearly one of dimensions in the city dynamics. In this work, we propose a technique called city image and show its applicability taking as examples eight different cities. The resulting image is a way of summarizing the city dynamics based on transition graphs, which map the movements of individuals in a PSN. This technique is a promising way to better understand the city dynamics, helping us to visualize the invisible images of cities.