Image Vector Quantization Indices Recovery using Lagrange Interpolation (original) (raw)
Related papers
Image Vector Quantization codec indexes filtering
Serbian Journal of Electrical Engineering, 2012
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.
Image Vector Quantization for Interframe Coding Applications
In this paper, an interframe image coding via vector quantization (VQ) is considered. Here, we identify the moving vectors in each frame using a block matching technique. A prediction is estimated by searching in the direction of minimum distortion in the previous reconstructed frame, then the differential vectors are quantized via a modified VQ to achieve high compression. Simulations have shown that an average PSNR=33 can be achieved at a bit rate R=0.15-0.25 bpp.
Signal Processing, 2004
In this paper, new vector quantization (VQ) based techniques for reducing the effect of channel noise in image transmission are introduced. In one of the proposed methods, i.e., component VQ (CVQ), by transmitting the components of code-vectors (and not their indices), we will be able to reduce the effect of channel noise. To decrease the output rate of CVQ, a modification of CVQ called sub-index VQ (SIVQ) is introduced. This method has errorcorrecting capability as well. Simulation results show that CVQ and SIVQ techniques are robust to channel errors and surpass the conventional method, in which the indices of the selected code-vectors are protected using a Reed-Solomon code. An error concealment technique called shifted block concealment (SBC) has been also introduced. The stationarity assumption of small blocks of images has been utilized to conceal losses in this method.
A new efficient image compression technique with index-matching vector quantization
IEEE Transactions on Consumer Electronics, 1997
A new efficient image compression technique is presentmecl for low cost, applications, such as multimedia. arid videoconferencing. Since address vector quantization (A-VQ) , proposed by Nasrabadi and Feng for image coding, has t,he main disadvantage of high computational complexity of reordering the address codebook at the transmitter and t,he receiver during encoding of each block, we propose a new efficient approach to overcome this disadvantage. The proposed algorithm is based on tree search vector quantizalioii via. multi-path search and index matching in index codebook, and may achieve a better performance as well as alphabet. Z = { I , 2 ,. .. , N } denote the finitme index alphabet, and A denote the finite reproduct,ion alphahet. We assume A c A. Let. E Ak be an input. vector sour(-e, a.nd let, C = {Cl, C'z,. .. , c."} be a. finite codebook cont,a.ining N codevectors, where N = 2"? R > 0, and for each 1 5 i 5 N, Ci is called the codewmrd or teinplak. A k-D vect,or quantizer Q with rate R is a, mapping Q : A k + C such that, low computational complexity. We theoretically prove t1ia.t. the proposed algorithm is superior to the A-VQ algorithm a.nd experimentally show that a lower bit. rate tha.n that, of where 8 is the encoder, 2, is the decoder, and Ri is the it ,li Voronoi cell with the centroid Ci defined by the .4-VQ is obtained. I. INTRODUCTION Clearly Recently, the topic of d a h compression (or source coding) especially for image and video, ha,s become attra.ctive clue to the demands in some applications, such as videoconf'erencing and multimedia. Vector quantiza.tion (VQ) has heen found to be an efficient coding technique due to i1.s inherent abilit,y to exploit the high correlation between the neighboring pixels. Some excellent survey articles and hooks are given in [l] [a]. Essentially, VQ coding technique can be viewed as a pattern matching method. VQ is a, block coding procedure by which blocks of k samples from a. given data, source are approximated by vector pa.tterns or t,einplat,es from a. set, of code vectors, commonly called a. codebook. VQ is widely used in image/video and speech compression applications, because simple table look-up encoding arid decoding procedures may be used. In this paper, 2-dimensional (2-D) informat.ion source is considered since a 2-D raster scan of the image is adopted. Let d denote the nonempty finite discrete source U Ri = Ak and R, n R, = (d if i # j. Here, k = 4 x 4 = 16 since 4 x 4 block c.otling is assumed. During the encoding of a digit,al image, t.he best, possible match (mininiuin dist,ortion, e.g., miniiriuni Euclidean distance) is extracted t,o represent the input vect,or. The codeword index i , i E Z = { 1 , 2 ,. .. , N } , is then t,ransinitted to the receiver where index i is decoded by a simple t,able look-up decoding process. Act,ually, the codebook is the key part of the vector quantizer. There are several different, approaches to the codebook design. A popular and well-known codebook design procedure. proposed by Linde. B w o , and Gray (LBG) [3], is a geiieralized (or vect,or) version of the Lloyd clustering algorithm for a scalar quantizer. 1n t.he st,andard rriemoryless VQ tmeclinique. t,he pixel (or intrablock) correlation is esploit.ed but. the int.erblock c-orrelat.ion is totally ignored. The interblock correla.t,ioii
Index compressed image adaptive vector quantisation
Signal Processing-image Communication, 1996
This paper introduces an improved image adaptive vector quantisation technique-index compressed image adaptive vector quantisation (IC-IAVQ). Despite its advantage over the universal codebook VQ, basic image adaptive VQ (IAVQ) is still suboptimum; it neglects the correlation among block indices in the encoded image. The new technique, IC-IAVQ, overcomes this suboptimality through a pre-processing and lossless compression of block indices. Simulation results using several images show that IC-IAVQ outperforms IAVQ and entropy coded IAVQ, especially at low bit-rates by about 2dB on average.
Interleaved reception method for restored vector quantization image
TELKOMNIKA (Telecommunication Computing Electronics and Control), 2016
The transmission of image compressed by vector quantization produce wrong blocks in received image which are completely different to the original one which makes the restoration process too difficult because we don't have any information about the original block. As a solution we propose a transmission technique that save the majority of pixels in the original block by building new blocks doesn't contain neighborhood pixels from the original block which increase the probability of restoration. Our proposition is based on decomposition and interleaving. For the simulation we use a binary symmetric channel with different BER and in the restoration process we use simple median filter just to check the efficiency of proposed approach.
Anais do XVIII Simpósio Brasileiro de Telecomunicações
A serious problem related to the transmission of vectorquantized images through noisy channels is that whenever errors occur, the nearest neighbor rule is broken. As a consequence, very annoying blocking e ects may appear in the reconstructed images. In the present work, a simple and fast method for organizing the vector quantization (VQ) codebooks is presented. The key idea behind the proposed method is to ensure that similar (dissimilar) binary representations of the indexes of the codevectors correspond to similar (dissimilar) codevectors themselves. It is shown that the organized codebooks improve the performance of the transmission system in the sense that they lead to reconstructed images with better quality when compared to the ones obtained by using non-organized codebooks.
“Survey on Vector Quantization for Image Compression”
2020
This paper presents a hybrid (loss less and lossy) technique for image vector quantization. The codebook is generated in two steps and first step is training set is sorted based on the magnitudes of the training vectors and step 2 is from the sorted list, training vector from every nth position is selected to for the code vectors. Followed by that, centroid computation with clustering is done by repeated iterations to improve the optimality of the codebook. The code book thus generated is compressed (Iossy) to reduce the memory needed to store the codebook with the slight degradation in the quality of the reconstructed image. The future wireless networks, such as Centralized Radio Access Network (C-RAN), c will requirement to deliver data rate about 100 times to 1000 times the current 4G technology. For C-RAN based network layout, there is a pressing The future wireless networks, such as Centralized Radio Access Network (CRAN), will need to deliver data rate about 100 times to 1000 ...
Modified Vector Quantization Method for Image Compression
Transactions On Engineering, Computing …, 2006
A low bit rate still image compression scheme by compressing the indices of Vector Quantization (VQ) and generating residual codebook is proposed. The indices of VQ are compressed by exploiting correlation among image blocks, which reduces the bit per index. A residual codebook similar to VQ codebook is generated that represents the distortion produced in VQ. Using this residual codebook the distortion in the reconstructed image is removed, thereby increasing the image quality. Our scheme combines these two methods. Experimental results on standard image Lena show that our scheme can give a reconstructed image with a PSNR value of 31.6 db at 0.396 bits per pixel. Our scheme is also faster than the existing VQ variants.