The most challenging issue of the navigation assistive systems for the visually impaired is the instantaneous and accurate spatial localization of the user. Most of the previous proposed systems are based on GPS sensors. But, low cost versions have clearly insufficient accuracy for pedestrian use. Furthermore, they are confined to outdoor navigation with severe failing in urban area. This paper presents a new approach for localizing a person by using a single body mounted camera and computer vision techniques. Instantaneous accurate localization and heading estimates of the person are computed from images as the trip progresses along a memorised path. A first portable prototype has been tested for outdoor as well as indoor pedestrian trips. Experimental results demonstrate the effectiveness of the vision based localization: the accuracy around twenty centimetres, allows guiding and keeping the blind in a navigation corridor along the intended path. In combination with a suitable guiding interface, such a localization system will propose a convenient assistive navigation for the visually impaired.