The main objective of the thesis is to find a method that permits to insert a quantity importing information in a sequence video without damaging it visually. The accent has been put therefore on the compromise to insert the maximum of bits in the video and to keep the invisible mark. On the other hand, he/it is not interesting in the practice to get an algorithm that would be too sensitive to the attacks. We are going to propose a method of based picture tattoo on the networks of ondelette. The main innovation of our method consists in using a widely-known technique in artificial intelligence the networks of ondelettes and to apply it at the domain of the tattoo of stationary pictures. The thus proposed method will be applied in the tattoo video for two different applications: The protection of the rights of authors and the indexing.