The Study Of Some Fractal Moving Image Compression Methods

Fractal moving image (video) compression was firstly proposed by M. F. Barnsley after he used the fractal to the still image compression and made some extension. For the development of modern communication technology problems about a great deal of storage, record and transmission for varied still and moving images are proposed, which supply for the real significance for the image compression. Based on it in my thesis I have done a serial of research work on the fractal moving image compression and achieved some results. Following is specific work: A novel algorithm for fractal video sequence coding based on the extended quadtree video partition with the circular prediction mapping and the noncontractive mapping is proposed. The proposed algorithm can effectively exploit the temporal correlation in real image sequences, since each range block is approximated by the domain block in the adjacent frame, which is of the same size as the range block. Moreover, I present a method of interpolation between two isolated frames which give a new thinking way for the improvement of compression ratio in low bit rate transmission. Computer experiment proves the results are relatively satisfactory. What is focused is the fractal approximation of multidimensional function in L2 . In particular, we treat a position-dependent approximation with no search using orthogonalbases of L . We describe a framework that establishes a connection between the classicorthogonal approximation and the fractal approximation. From the computational perspective, we can avoid to solve linear systems often suffering from ill conditioning, which was needed in former fractal approximation techniques. Moreover, using orthogonal bases we obtain the most compact representation of approximation. As a direct application we show some results on the compression of gray scale digital images. Based on the two-dimensional DCT a three-dimensional DCT (discrete cosine transformation), which transforms 3D video data from special-temporal field into frequency field is extended. Then using fractal we seek a most approximated range block for each variable domain block in frequency field. Experiment proves it has some application perspective in non-real time video procession. An efficient scheme for video coding is presented which utilizes progressive fractal coding, called wavelet-based fractal approximation (WBFA), and motion compensation(MC). In the scheme, the MC errors frames are encoded by the progressive fractal coding. The key of the scheme is the progressive fractal in wavelet field, which irregularly samples in fine wavelet field. Experimental result shows the scheme is more suitable for low bit rate video coding.