The thesis describes usage and generation of motion information in the field of video coding. Hierarchical structures for computation of this motion information are suggested. These hierarchical structures can be used also for other applications.

One part of the motion information consists of vectors indicating the displacement between successive frames in an image sequence. These vectors are frequently used in existing video coding schemes. Algorithms for calculating the vectors are described and analyzed, and it is shown how the computational complexity can be lowered for some of the algorithms.

The other part of the motion information is local certainty, which gives an indication about the probability that the vector corresponds to the true displacement. The certainty part of the motion estimation has not often been utilized in the field of video coding. In the thesis, different certainty measures are evaluated. Two applications of the certainty measures are examined: pre-filtering for noise reduction, and reduction of the required number of bits to represent the video sequence in a scalable wavelet coding scheme.