Statistic approach of vector quantization uses code-books of rounded vectors which allow quasi-optimal coding for a given rate. Therefore these code-books have no structure and require big memory size. Regular lattices of points give the possibility of generating a great number of points from a short number of vectors. This can permit us to solve the problems of the statistic approach. However the lattices presents their own problem, effectively they are only applied for uniform distribution sources. In order to avoid these two kinds of problems, another approach is investigated; It consists of designing a new quantizer by combining the two techniques. In this paper, we present this approach which we call hybrid; first few statistic vectors of the source are performed by using the LBG (Linde, Buzo and Gray) algorithm [2], then each of them leads to a Gosset lattice. The results obtained in image coding are also presented.
Djouadi, M., & Berkani, D. (1999). IMAGE CODING USING HYBRID VECTOR QUANTIZATION. The International Conference on Electrical Engineering, 2(2nd International Conference on Electrical Engineering ICEENG 1999), 699-706. doi: 10.21608/iceeng.1999.62595
MLA
M. S Djouadi; D. Berkani. "IMAGE CODING USING HYBRID VECTOR QUANTIZATION", The International Conference on Electrical Engineering, 2, 2nd International Conference on Electrical Engineering ICEENG 1999, 1999, 699-706. doi: 10.21608/iceeng.1999.62595
HARVARD
Djouadi, M., Berkani, D. (1999). 'IMAGE CODING USING HYBRID VECTOR QUANTIZATION', The International Conference on Electrical Engineering, 2(2nd International Conference on Electrical Engineering ICEENG 1999), pp. 699-706. doi: 10.21608/iceeng.1999.62595
VANCOUVER
Djouadi, M., Berkani, D. IMAGE CODING USING HYBRID VECTOR QUANTIZATION. The International Conference on Electrical Engineering, 1999; 2(2nd International Conference on Electrical Engineering ICEENG 1999): 699-706. doi: 10.21608/iceeng.1999.62595
)
View on SCiNiTO