Modern information technology / 2. Computer Science and Programming

Lishchuk O.O.

Vinnitsa National Technical University, Ukraine

Analysis of the base methods of coding images

The basic method of coding digital image source is pulse-code modulation (PCM). It is characterized by the fact that each encoded word in digital form corresponding to quantized in time and amplitude video countdown. This should be carried out sampling theorem requirements fd> 2W0, where W0 maximum frequency contained in the signal. To prevent false contours on monochrome image should be 50 quantization levels corresponding to 6-8 bit code word for each element (pixel) image. Because of the large volumes of information applies only when PCM vnutristudiyniy parallel transmission of television images and code as the baseline canonical representation of images in digital form [1,2,3,4].

Among the methods of coding predictive most studied differential pulse code modulation (DPCM). The essence of this DPCM. Organized prediction brightness of each image next item based on a linear combination of brightness values of previous elements. Evaluation of the resulting prediction is subtracted from the true brightness image element and quantized difference signal to a small number of levels. Due to this and achieved the reduction of data [1,3,7].

The methods of image coding predictions allow compress images in 2-2,5 times in a simple technical implementation. Among the drawbacks of these methods should include:

• errors in the field of extreme brightness;

• low immunity.

Interpolation methods based on numerical approximation methods by which a sequence or two-dimensional array of samples submitted brightness approximately by continuous functions [8,9]. Thus only a few samples encoded image and neighboring obtained by interpolation polynomials, of course, no more than the third degree. The compression ratio, which is achieved by using these methods is 5-6.

The main drawback of interpolation methods is a large amount of computation in the interpolation polynomials of high degrees, as well as the need to keep basic coordinate readout image.

An important class image coding methods - a method based on coding changes. Encoding based on change - indirect method. Picture subjected unitary mathematical transformation, resulting quantized transform coefficients and encoded for transmission over a communication channel or to a file. Most image conversion factors mentioned many relatively small. These factors can often reject or allocate them to encode a small number of bits. Most often two-dimensional orthogonal transformation. It's like Karhunen-Loeve, discrete Fourier transform (DFT), discrete cosine transform (DCT), Walsh-Hadamard conversion [2,3,8,9] and others. The compression ratio when using these methods may reach 6-8, so some of these methods (SCE) are widely used as a basis for industry standards of image coding. The common drawback of these methods is quite high computational complexity, and inability to resolve a number of image processing tasks to the reduced volume of data.

Among the statistical methods most widely used methods of image coding block. Blocks of size MxN elements encoded according to the probability of their occurrence. For most probable configurations use short code words, and for less probable - long code words (Huffman algorithm), resulting in data compression is achieved. The compression ratio when using these methods may reach 4-5.

Promising to encode both moving and still images are component-wise coding method. The feature of this method is the formation of multiple two-dimensional signals carry information about the details of images of different sizes. The presence of multiple channels separate image processing parts of different sizes, can effectively encode the image as intraframe and interframe methods allowing for the perception of visual information analyzer man. An important advantage of this method is that the data presentation formats compression are components of the original image that visually represents each component image with a certain degree of resolution, which in turn allows you to solve a series of image processing tasks to the reduced volume data.

Although almost achieved compression ratio when using this method slightly lower compared with the methods of transformation coding, its technical implementation is much simpler and therefore much greater speed.

To compress the images in real time on set parameters such as ease of technical implementation, computational complexity, compression ratio best results are obtained when using Wavelet-coding based on pyramid schemes decomposition of the original image into components (S-transformation) [6]. This algorithm is aimed at compressing color and black and white images with smooth transitions. Ideal for pictures of type X-ray images. The compression ratio in the range of 5-100. At high compression ratio for sharp borders, especially diagonal, possible distortion. An important advantage of the S-transformation is the ability to show "coarsened" images (low resolution) using just the beginning of the file.

The highest compression ratio provides a method of fractal image compression [5] opened in 1988. Fractal compression process is based on the assumption that the image of the real world with affine redundancy. The compression ratio can reach 50-60 times. The main disadvantage of this method is the large computational complexity. However, given the high rate of compression, which can be obtained by this method, and a huge progress in increasing performance of microprocessors and other hardware should be expected of the wide application of this method in the next few years.

Literature

1. Digital TV / Krivosheev MI, Vilenchik SL, Krasnosel'skii IN and etc.; Ed. MI Krivosheeva. - M .: Communication, 1980. - 264 p.

2. Digitizing television images / Ed. I.I.Tsukkermana. - M .: Radio and Communications, 1981. -240 p.

3. M. Ptacek Digital TV. Theory and Technology: Per. with the Czech. - M .: Radio and Communications, 1990. - 528 p.

4. CCIR. Digital encoding of video images in the studio. Recommendation 601. The 14th Plenary Assembly (Dubrovnik, 1986).

5. M. Barnsley, Anson L. Fractal image compression // World PK.¬¬-1992. - № 10. - C. 52-58.

6. Vatolin DS Image compression algorithms. - M .: Publishing Department of the Faculty of Computational Mathematics and Cybernetics, Moscow State University. MSU (LR license number 040777 of 23/07/96), 1999 - 76.

7. Kharatishvili NG Digital predictive coding of continuous signals. - M .: Radio and Communications, 1986. - P. 3-25.

8. W. Pratt Digital Image Processing: Ed. from English. - M .: Mir, 1982. - V.2. - 480 with.

9. Babak is the іn. Obrobka signalіv /V.P.Babak, VS Handetsky, E. Shryufer. - K .: Lybid, 1996. - 392 p.