A B C D E F G H I J K L M N O P Q R S T U V W X Y Z All
Vimala, S.
- Codebook Generation for Vector Quantization by Sorting the Sum of Sub Vectors
Authors
1 Department of Computer Science and Applications, Gandhigram Rural Institute, Gandhigram-624302, Tamilnadu, IN
2 Department of Computer Science, Mother Teresa Women’s University, Kodaikanal-624102, Tamilnadu, IN
Source
Digital Image Processing, Vol 2, No 8 (2010), Pagination: 266-271Abstract
Vector Quantization is a lossy image compression technique. In this paper, we propose a novel idea for generating a codebook by sorting the sum of sub vectors. The training vectors are sub divided into four sub vectors each consisting of four elements.The sum of sub vectors 2 and 4 are subtracted from the sum of sub vectors 1 and 2. The resultant values are used to sort the training vectors. From the sorted list, the training vectors at every nth position are selected to form the codebook. The experimental results and the comparisons show that this method gives better performance with respect to the time taken to generate the codebook and the PSNR value (quality of the reconstructed images). The computational complexity involved is also very less. The codebook generated using the proposed method is optimized using the iterative clustering method. The quality of the reconstructed image is improved to a significant value.
Keywords
Image Compression, Sub Vector, Training Vector, Code Vector and Codebook.- Codebook Generation for Vector Quantization with Edge Features
Authors
1 Department of Computer Science and Applications, Gadhigram Rural Institute, Gandhigram – 624302, Tamil Nadu, IN
2 Department of Computer Science, Mother Teresa Women’s University, Kodaikanal – 624 102, Tamil Nadu, IN
Source
Digital Image Processing, Vol 2, No 7 (2010), Pagination: 194-198Abstract
In this paper, we propose a new technique to retain the edges of the images while compressing them. The image is divided into small blocks of size 4 x 4 pixels. From the list of N image blocks, M (M < N and M is the desired size of codebook) representative blocks are selected to form the codebook. In ordinary codebook generation, all the image blocks are treated equally. But, in the proposed method, the image blocks are classified into high detail blocks and low detail blocks. The high detail blocks represent the edges and the low detail blocks represent the shaded blocks. The high detail blocks are given preference while generating the codebook. Categorization of image blocks improves the quality of the reconstructed images. Codebooks of sizes (M) 128 and 256 are created with the images of size 256 x 256 pixels. The proposed method gives better performance when compared with the results of few existing techniques.