Abstract
The principal objective of this thesis is to unify disparate 3D information and then realize scalable visualization in a client/server environment that is heterogeneous in terms of network, computing and memory resources. For scalability we are exploiting the multiresolution nature of the discrete wavelet transform (DWT) from the state of the art JPEG2000 codec. The data unification is being carried out through DWT domain blind data hiding that may either be fully or adaptively synchronous. A typical surface based 3D visualization requires at least two sets of data: a 2D intensity image, called texture, with a corresponding 3D shape rendered in the form of a range image, a shaded 3D model and/or a mesh of points. A range image, also sometimes called a depth image, is an image in which the pixel value reflects the distance from the sensor to the imaged surface. The texture is a corresponding 2D color image which is overlaid onto a model produced from the depth map by triangulation. For data hiding, the range data is first subjected to DWT while the texture data is input to the JPEG2000 encoder.
The JPEG2000 coding pipeline of texture is interrupted at some stage after the DWT step and the DWT domain range coefficients are embedded in the all or a subset of texture subbands. The embedded data is reintroduced to the JPEG2000 pipeline at the same step where the interruption was made. Since the JPEG2000 format is conserved during the process, the resultant code can be sent across any communication channel like any other JPEG2000 file. The resolution scalability of wavelets and the synchronized character of our techniques enable a 3D visualization even with fewer than original resolution bands as a result of partial or delayed data transfer. The method thus enables to effect a visualization from a fraction of data in the form of the lowest subband, of a particular resolution level. In the first phase of this work the thrust was on the perceptual transparency and that is<br />why least significant bit (LSB) embedding was employed for both the synchronous and adaptively synchronous cases. The second phase concerns robustness and that is a why spread spectrum (SS) strategy is utilized in embedding. During the latter phase the imperceptibility has not been ignored at all as the embedding is removable. Examples from the two main areas of terrain and face visualization have been taken as case studies to prove the utility of our methods. In the third and final phase we take a broader aspect of the problem when we try to render the underlying tile components of a heterogeneous tessellation, seamlessly. This seamless joining of tiles is ensured through special DWT domain smoothing functions.