Image Vector Quantization codec indexes filtering
Vector Quantisation (VQ) is an efficient coding algorithm that has been widely used in the field of video and image coding, due to its fast decoding efficiency. However, the indexes of VQ are sometimes lost because of signal interference during the transmission. In this paper, we propose an...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Faculty of Technical Sciences in Cacak
2012-01-01
|
| Series: | Serbian Journal of Electrical Engineering |
| Subjects: | |
| Online Access: | http://www.doiserbia.nb.rs/img/doi/1451-4869/2012/1451-48691202263L.pdf |
| Summary: | Vector Quantisation (VQ) is an efficient coding algorithm that has been
widely used in the field of video and image coding, due to its fast decoding
efficiency. However, the indexes of VQ are sometimes lost because of signal
interference during the transmission. In this paper, we propose an efficient
estimation method to conceal and recover the lost indexes on the decoder
side, to avoid re-transmitting the whole image again. If the image or video
has the limitation of a period of validity, re-transmitting the data wastes
the resources of time and network bandwidth. Therefore, using the originally
received correct data to estimate and recover the lost data is efficient in
time-constrained situations, such as network conferencing or mobile
transmissions. In nature images, the pixels are correlated with their
neighbours and VQ partitions the image into sub-blocks and quantises them to
the indexes that are transmitted; the correlation between adjacent indexes is
very strong. There are two parts of the proposed method. The first is
pre-processing and the second is an estimation process. In pre-processing, we
modify the order of codevectors in the VQ codebook to increase the
correlation among the neighbouring vectors. We then use a special filtering
method in the estimation process. Using conventional VQ to compress the Lena
image and transmit it without any loss of index can achieve a PSNR of 30.429
dB on the decoder. The simulation results demonstrate that our method can
estimate the indexes to achieve PSNR values of 29.084 and 28.327 dB when the
loss rate is 0.5% and 1%, respectively. |
|---|---|
| ISSN: | 1451-4869 2217-7183 |