Problems of video compression providing large coefficient without noticeable deterioration in the visual quality of the decoded images are given in the article. To improve the efficiency of image encoding the method proposed real-time prior dimming image pixel source 2 times and restore their values during decoding. Experimental results on the effectiveness of this method for processing video are given. With development of digital TV broadcasting more and more requirements are imposed to quality support of images that requires increase in data transmission rate. Division of values of pixels into integers when coding and multiplication of the received results by the same numbers when decoding can be the simplest and effective method of the brightness transformation of the initial image.
Keywords: coefficient of compression of video data, efficiency of coding of images, processing of a video stream, coefficient of compression of the video codec on the veyvlet transformations,VT
В материалах доклада рассматриваются проблемы обеспечения больших коэффициентов сжатия видеоданных телевизионных изображений без заметного ухудшения визуального качества декодированных изображений. Для повышения эффективности кодирования изображений в реальном масштабе времени предложен метод предварительного уменьшения яркости пикселей исходных изображении в 2 раза и восстановление их значений при декодировании. Приводятся экспериментальные результаты по оценке эффективности применения данного метода для обработки видеопотока. С развитием цифрового телевизионного вещания все больше требований предъявляется к обеспечению высокого качества изображений, что требует увеличения скорости передачи данных. Наиболее простым и эффективным методом яркостного преобразования исходного изображения может быть деление значений пикселей на целые числа при кодировании и умножение полученных результатов на эти же числа при декодировании.
Ключевые слова: коэффициент сжатия видеоданных, эффективность кодирования изображений, обработка видеопотока, коэффициент сжатия видеокодека на вейвлет преобразованиях, ВП
With development of digital TV broadcasting more and more requirements are imposed to quality support of images that requires increase in data transmission rate. But the frequency resource is limited and the board for its use is rather high. Therefore for quality support a body of transmissions in case of low data transmission rates, more effective methods of a compression of video data are required. At the same time, usually compression of video data is made due to elimination of different types of the redundant information in images (code, structural, statistical, psycho visual, temporal). However, the problem is that value of the redundant information in images strongly depends on structure of the video topic. So images with rather uniform background contain the considerable volume of the redundant information and therefore contract well, and images with fine details and the considerable overfalls of brightness contain not enough redundancies and respectively contract badly.
Besides in television efficiency of coding is influenced by the volume of office information (metadata) which usually arises at block division of images for elimination of interpersonal redundancy on the basis of compensation of the movement. At the same time for more exact compensation of the movement of video objects blocks of the smaller size are required, and it leads to increase in their number and respectively to increase in the massif of metadata which are added to the coded video data of the compressed image. And it leads to reduction of resultant size of compression as metadata are transferred in a uniform stream with video data. So, when processing images of 720х576 in size blocks of 16х16 pixels the massif of metadata, at an average speed of the movement of objects, occupies 3240 bytes. At average coefficients of compression of video data (40–50 times) such makeweight not strongly influences resultant coefficient of compression. However at compression in 100 and more times such massif already becomes considerable.
One more factor influencing efficiency of compression of video data is brightness of the initial image as the size of values of pixels influences dimension of a final code. Thus, reduction of brightness of the image before coding and restoration it when decoding allows to reduce the required number of bits, for transfer of values of pixels and respectively data array of an output stream. Such kind of approach is used in the JPEG-2000 standard, however, in JPEG-2000 from values of pixels their average value on a shot is taken away. It brings to that on plots with big differences of brightness of small videos of objects average value can be quite low and the prize will be insignificant. [1]
Division of values of pixels into integers when coding and multiplication of the received results by the same numbers when decoding can be the simplest and effective method of the brightness transformation of the initial image. This method is very simple in realization, has high speed and doesn't form additional metadata. But at operations of division there are fractional numbers which when rounding to integer values cause irrevocable loss of information that respectively leads to distortion of a color rendition of the image at restoration. However, if it is limited to decrease in brightness of images twice, then the error of rounding won't exceed the size of the younger category that at 8 bit coding is visually almost imperceptible.
For an assessment of influence of decrease in brightness of initial images twice on coefficient of compression of the video codec on the veyvlet transformations (VT) the research on processing of 4 test images of various plots and genres without use of the quantumer of the video codec has been conducted.
When carrying out experiments initial images, then the same images from brightness (pic. 1) reduced twice then when decoding there was a recovery of the processed images contracted in the beginning.
Fig. 1. Initial and with the reduced brightness the test image «Snow-storm»
The initial and decoded image of a video topic «Snow-storm» are submitted in pic. 2, and experimental results of processing of images are reduced in table 1, and presented in the form of histograms of dependences of coefficient of compression of the initial and transformed images (pic. 3) and a prize percentage of application of this method of the brightness transformation (pic. 4).
Fig. 2. The initial and restored image of a video topic «Snow-storm»
|
№ |
Video Topic Type |
Compression Coefficient |
Profit% |
|
|
initial |
transformed |
|||
|
1. |
Snow-storm |
8,4 |
13,1 |
56 |
|
2. |
Sunset |
28 |
42 |
50 |
|
3. |
Girl |
23,8 |
35,7 |
50 |
|
4. |
Nature |
8,5 |
13,3 |
56,5 |
Fig. 3. Coefficient of compression of the initial and transformed images
Fig. 4. Results of a prize on compression of images due to the brightness transformation twice
Apparently from results of researches, visual quality of the initial and restored images submitted in fig. 2 almost not distinguishable. At the same time decrease in brightness of the image gives approximately 50 percentage prize on compression of video data twice. Thus this method of preliminary processing of TV of images allows to increase significantly overall performance veyvlet the video codec practically without deterioration in visual quality of images and preservation of speed of their processing.
References:
- Gavrilov I. A., Rakhimov T. G., Puziy A. N., Nosirov H. H., Kadirov Sh. M. «Digital television», publishing house of " Тоp Image Media " Tashkent 2016 400p
