The invention relates to a method for inputting a text in a distributed system that comprises a secured device, or server, communicating with at least one electronic device, or terminal. The method makes it possible to prepare an image by associating certain values to areas defining said image. The image is projected to the user who can touch the areas of the images that he wants to select as he would do with the keys of keyboard. The finger movements are analysed and the position of the virtual keys selected by the user is transmitted to the server that establishes a correlation between these positions and the previously associated values.