Vector quantization in subband coding of images
Guardado en:
| Publicado en: | ProQuest Dissertations and Theses (1995) |
|---|---|
| Autor principal: | |
| Publicado: |
ProQuest Dissertations & Theses
|
| Materias: | |
| Acceso en línea: | Citation/Abstract Full Text - PDF |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Resumen: | Subband coding is an effective method for image compression. It is based on the decomposition of the original image into multiple, nearly uncorrelated, sub-images using a perfect reconstruction filter bank. In general, each sub-image is quantized individually and subsequently recombined with the other subbands. High compression is achieved by properly allocating the desired bit rate among the subbands. Another significant achievement in signal compression is vector quantization. This lossy technique divides the signal into vectors instead of scalars where each vector is approximated from a library of representative vectors, called a codebook, based on a defined distortion criterion. Vector quantization achieves higher coding gain than scalar quantization by properly exploiting the statistical dependencies of the signal. This thesis focuses on the integration of vector quantization in subband coding of images. A method is presented which applies vector quantization in each of the high frequency subbands. A vector-based bit allocation procedure is proposed that distributes the available bits among the different regions of the subbands. This is achieved according to the activity levels in each region. A unitary transformation is applied to each subband to take advantage of the pronounced directional activity in each subband. Vector quantization is used in the transform domain where variable sized sub-vectors are constructed according to the associated subband and the total bit rate of the image. Finally, a multi-codebook vector quantization method is designed for each subband where codebook selection is based on information from lower level subbands. This method exploits the noticeable dependency among multiple level subbands. |
|---|---|
| ISBN: | 9798208800034 |
| Fuente: | ProQuest Dissertations & Theses Global |