The underwater cartography has made great progress in the last decade. In this paper, we discuss of the 3D underwater cartography problem and propose a multimodal fusion approach. The work presented in this paper is about the analyze and 3D reconstruction of archeological objects. Using an uncalibrated single camera, we propose to describe a method to firstly, calibrate this camera integrating variation of the water optical refractive index to the distortion model. And secondly, robust and stable features have been extracted. A comparison of two methods of detecting and matching points is presented. Finally, we estimate the movement between different images and compare two methods.