Short Essay of MATLAB Vector Quantization

               This technique is used to attain the Compact representation of an image is attaining without much data loss is referred to as image compression along with all vital information. Into two types lossy image compression could be categorized, namely as SQ (scalar Quantization) as well as VQ (Vector Quantization. For the last two decades for image source coding as well as data, speech vector quantization has been one of the most competent methods. Using some distortion measure vector quantization consists of processing the input samples in groups into a group of well-defined vectors. For image compression vector quantization is a technique is discussed here.

           Kaukoranta et al., [KAUK00] suggested that using GLA algorithm the vector Quantization technique could be exploited for its ease of use in image compression. In two conditions the criteria of diminishing the Distortion the Lioyed algorithm: to diminish the squared error distortion |x-x(c)|^2 as the x is specified as the input vector. The 2nd condition is that the c is known the code,  x^' (c) can be calculated by the input x but firstly it should be satisfied the first condition. In batch mode the machine is carried out with respect to the first criterion, to utilize the vector quantization, as well as with respect to the 2nd condition the decoder is optimizing, until a minimum distortion is achieved. With different initial code vectors to execute the algorithm many times, it might be important, In order to avoid the local minimal difficulty. (Kaukoranta & al, 2015)

            V.K. Jain et al. [JAIN96] According to Jain, for image compression he explained the Entropy-Constrained vector quantization method. In image compression, this method incorporates the concept of an activity map. In the vector subband coding is the fundamental issue compelling pressure of the satisfactory vitality though perceptually significant, upper-groups of the decompressed pictures. In the upper sub-groups as the components of the critical vectors show the shape data of the unique pictures that must be compacted, therefore, the position, what's more, the introduction of such edge points of interest could be resolved from the unique picture itself. To encode the lower index shorter symbols were allocated, however, the profitable is very high.  In the sending of indices in the state code book, the code words have the different cost. In general, in the coding process, the adjacent code vectors are preferred constantly. To generate the active map this information could be utilized, to obtain transmission control maps which in turn is used. The transmission of vector turn on/off the control maps is used. This method is also applied to color image coding. (V.k.Jain & Boopathy, 2010)

            F. Medeiros et al. [MADE00] suggested a technique named Wavelet-Based Vector Quantization for quantizing as well as compressing images. In this technique, the multiple stage discrete code words of wavelet transform were considered as well as in both examinations exploited along with processes of design for image compression. For each stage, the major characteristics feature of this technique is discovering representative code vectors that are found in k groups by initially integrating n code words, codebook size gives them in. After attaining the k groups the centroids of the groups are considered as the representative vectors. (F. Madeiro & al, 2009)

                Multi-vector Quantization is one of the VQ methods that is proposed by Noritaka Shigeru et al. [NORI05] to compress a picture to construct multiple independent codebooks. From the multiple codebooks, low-quality picture restores through MVQ, as well as then it is useful to convert the low-quality picture into the high-quality picture. To overcome the inefficiency in compression rate it exploits an efficient coding method. When compression rate is less than the certain value then MVQ perform better than other VQ technique.

1.                      The vector Quantizer Encoder object performs vector quantization encoding. By computing a distortion based on Euclidean the object finds the nearest code word. We should learn two things when using vector quantization encoding:Generate the dsp. Vector Quantizer Encoder purpose as well as set its properties.

2.   With arguments call the object, as if it were a function.  (Mathworks, n.d.)`

Now we describe the syntax of the vector Quantization which we used in the Matlab

Reference of MATLAB Vector Quantization